RankingGPT-5.5
GPT-5.5
GPT-5.5 by OpenAI is a gpt-5.5 model. This profile collects the published benchmark results and reported effort settings for this model. Configurations are compared on the effort page. Ranks are on the leaderboard.
112 published benchmark measures · 158 measured effort results · 23 benchmark families contribute across 6 task areas. 5 capability estimates available in the full profile. See all results ↓ · Compare published benchmarks →
Model evidence summary
This profile combines published settings. It is not a runnable configuration or a leaderboard rank. Compare measured configurations →
Performance profile
Capabilities
Filled points are supported; hollow points are preliminary. Lines stop at unknown capabilities. Exact values and sources follow below.
Scores estimate outcomes against a shared reference panel; they are not accuracy percentages. Sparse or disconnected evidence cannot qualify an overall profile. Open a capability to inspect its evidence.
Reported effort · XHigh + unspecified
Settings reported in this model's published benchmark results, including results outside the aggregate. Effort names are provider-specific. These are not API defaults or equal compute budgets.
- XHigh: 49 observations
- Not specified: 106 observations
Mixed settings means multiple settings occur in the evidence. Best across efforts means the source selected its best reported result across settings; it does not mean Max. Unspecified settings stay unknown. This model-summary chart combines reported settings. The leaderboard keeps identified configurations separate and excludes unknown effort.
Inspect each result and its source ↓ · Download effort evidenceScore contributions and missing evidence
23 contributing families across 6 capabilities. Fixed reference panels do not change when the catalog expands.
Capability is fitted jointly across families. Capability estimates below describe different task areas; their weighted sum is not the Capability score.
Results without reviewed compatibility or a reference match remain in the raw evidence below. Coverage counts only contributing results.
Model information & shareable badge
- Lab
- OpenAI
- Catalog status
- active
- Availability
- Public provider catalog; account and region restrictions may apply
- Family
- gpt-5.5
- Released
- —
- Context
- —
- License
- proprietary
- Model card
- https://developers.openai.com/api/docs/models/all
- Default Capability family coverage
/badge/gpt-5.5.svg
GPT-5.5 measured results by effort setting
Each row is one published result for one effort setting. Evidence collected 2026-11-22. Compare the effort settings →
| Benchmark | Effort | Score | Source |
|---|---|---|---|
| AA-AnalystAgent | Non-reasoning | 7.5% | Artificial Analysis |
| AA-AnalystAgent | XHigh | 50.0% | Artificial Analysis |
| AA-Briefcase Elo | High | 1087 | Artificial Analysis |
| AA-Briefcase Elo | Medium | 1000 | Artificial Analysis |
| AA-Briefcase Elo | XHigh | 1135 | Artificial Analysis |
| AA-LCR v1.1 | High | 84.3% | Artificial Analysis |
| AA-LCR v1.1 | Low | 81.0% | Artificial Analysis |
| AA-LCR v1.1 | Medium | 83.0% | Artificial Analysis |
| AA-LCR v1.1 | Non-reasoning | 64.0% | Artificial Analysis |
| AA-LCR v1.1 | XHigh | 84.3% | Artificial Analysis |
| Agents' Last Exam | High | 23.0% | Agents' Last Exam |
| Agents' Last Exam | High | 21.7% | Agents' Last Exam |
| Agents' Last Exam | High | 19.7% | Agents' Last Exam |
| Agents' Last Exam | Medium | 21.4% | Agents' Last Exam |
| Agents' Last Exam | XHigh | 26.6% | Agents' Last Exam |
| APEX-Agents-AA | XHigh | 37.7% | Artificial Analysis |
| ARC-AGI-2 | High | 83.3% | ARC Prize |
| ARC-AGI-2 | Low | 33.3% | ARC Prize |
| ARC-AGI-2 | Medium | 70.4% | ARC Prize |
| ARC-AGI-2 | XHigh | 85.0% | ARC Prize |
| AutomationBench-AA | High | 44.3% | Artificial Analysis |
| AutomationBench-AA | Medium | 40.1% | Artificial Analysis |
| AutomationBench-AA | XHigh | 47.3% | Artificial Analysis |
| CritPt | High | 25.4% | Artificial Analysis |
| CritPt | Low | 8.0% | Artificial Analysis |
| CritPt | Medium | 18.6% | Artificial Analysis |
| CritPt | Non-reasoning | 1.4% | Artificial Analysis |
| CritPt | XHigh | 27.1% | Artificial Analysis |
| DeepSWE v1.1 | High | 64.4% | Datacurve |
| DeepSWE v1.1 | Low | 27.0% | Datacurve |
| DeepSWE v1.1 | Medium | 54.0% | Datacurve |
| DeepSWE v1.1 | XHigh | 67.0% | Datacurve |
| EnterpriseOps-Gym-AA | Medium | 44.1% | Artificial Analysis |
| EnterpriseOps-Gym-AA | XHigh | 46.6% | Artificial Analysis |
| Epoch · Chess Puzzles · v2.0.0 | Low | 30.0% | Epoch AI |
| Epoch · Chess Puzzles · v2.0.0 | None | 9.0% | Epoch AI |
| Epoch · Chess Puzzles · v2.0.0 | XHigh | 63.0% | Epoch AI |
| Epoch · FrontierMath Tier 4 · v2.0.0 | XHigh | 72.5% | Epoch AI |
| Epoch · FrontierMath Tiers 1–3 · v2.0.0 | XHigh | 85.3% | Epoch AI |
| Epoch · GPQA diamond · v1.0.11 | None | 77.3% | Epoch AI |
| Epoch · GPQA diamond · v1.0.8 | Low | 90.7% | Epoch AI |
| Epoch · Mystery Game Puzzles · v1.0.4 · Tool submission | High | 52.0% | Epoch AI |
| Epoch · Mystery Game Puzzles · v1.0.4 · Tool submission | Low | 28.0% | Epoch AI |
| Epoch · Mystery Game Puzzles · v1.0.4 · Tool submission | None | 18.0% | Epoch AI |
| Epoch · Mystery Game Puzzles · v1.0.4 · Tool submission | XHigh | 56.0% | Epoch AI |
| Epoch · OTIS Mock AIME 2024-2025 · v1.1.6 | Low | 84.4% | Epoch AI |
| Epoch · OTIS Mock AIME 2024-2025 · v1.1.6 | None | 57.8% | Epoch AI |
| Epoch · SimpleQA Verified · v1.2.0 | XHigh | 63.0% | Epoch AI |
| FrontierCode 1.1 · extended-codex | High | 54.0% | Cognition |
| FrontierCode 1.1 · extended-codex | Low | 45.6% | Cognition |
| FrontierCode 1.1 · extended-codex | Medium | 51.4% | Cognition |
| FrontierCode 1.1 · extended-codex | XHigh | 56.7% | Cognition |
| FrontierCode 1.1 · main-codex | High | 40.6% | Cognition |
| FrontierCode 1.1 · main-codex | Low | 30.6% | Cognition |
| FrontierCode 1.1 · main-codex | Medium | 36.6% | Cognition |
| FrontierCode 1.1 · main-codex | XHigh | 43.0% | Cognition |
| FrontierMath Erdős | XHigh | 0.0% | Epoch AI |
| GDP.pdf: All-pass Rate | High | 22.6% | Artificial Analysis |
| GDP.pdf: All-pass Rate | Medium | 20.6% | Artificial Analysis |
| GDP.pdf: All-pass Rate | XHigh | 21.2% | Artificial Analysis |
| GDPval-AA v2.1 Leaderboard | High | 1335 | Artificial Analysis |
| GDPval-AA v2.1 Leaderboard | Low | 1053 | Artificial Analysis |
| GDPval-AA v2.1 Leaderboard | Medium | 1242 | Artificial Analysis |
| GDPval-AA v2.1 Leaderboard | Non-reasoning | 978 | Artificial Analysis |
| GDPval-AA v2.1 Leaderboard | XHigh | 1357 | Artificial Analysis |
| GPQA Diamond | High | 93.2% | Artificial Analysis |
| GPQA Diamond | Low | 91.0% | Artificial Analysis |
| GPQA Diamond | Medium | 92.6% | Artificial Analysis |
| GPQA Diamond | Non-reasoning | 76.8% | Artificial Analysis |
| GPQA Diamond | XHigh | 93.5% | Artificial Analysis |
| Humanity's Last Exam | High | 45.0% | Artificial Analysis |
| Humanity's Last Exam | Low | 32.7% | Artificial Analysis |
| Humanity's Last Exam | Medium | 42.4% | Artificial Analysis |
| Humanity's Last Exam | Non-reasoning | 13.7% | Artificial Analysis |
| Humanity's Last Exam | XHigh | 45.8% | Artificial Analysis |
| IFBench | High | 71.6% | Artificial Analysis |
| IFBench | Low | 64.4% | Artificial Analysis |
| IFBench | Medium | 71.0% | Artificial Analysis |
| IFBench | Non-reasoning | 46.1% | Artificial Analysis |
| IFBench | XHigh | 75.9% | Artificial Analysis |
| ITBench-AA | XHigh | 45.8% | Artificial Analysis |
| LiveBench · AMPS Hard · 2026-06-25 | XHigh | 97.0% | LiveBench |
| LiveBench · code completion · 2026-06-25 | XHigh | 82.6% | LiveBench |
| LiveBench · code generation · 2026-06-25 | XHigh | 81.7% | LiveBench |
| LiveBench · connections · 2026-06-25 | XHigh | 100.0% | LiveBench |
| LiveBench · consecutive events · 2026-06-25 | XHigh | 88.8% | LiveBench |
| LiveBench · integrals with game · 2026-06-25 | XHigh | 99.0% | LiveBench |
| LiveBench · javascript · 2026-06-25 | XHigh | 63.6% | LiveBench |
| LiveBench · logic with navigation · 2026-06-25 | XHigh | 76.0% | LiveBench |
| LiveBench · math comp · 2026-06-25 | XHigh | 96.1% | LiveBench |
| LiveBench · olympiad · 2026-06-25 | XHigh | 91.4% | LiveBench |
| LiveBench · paraphrase · 2026-06-25 | XHigh | 71.7% | LiveBench |
| LiveBench · plot unscrambling · 2026-06-25 | XHigh | 74.1% | LiveBench |
| LiveBench · python · 2026-06-25 | XHigh | 55.0% | LiveBench |
| LiveBench · simplify · 2026-06-25 | XHigh | 66.3% | LiveBench |
| LiveBench · spatial · 2026-06-25 | XHigh | 98.0% | LiveBench |
| LiveBench · story generation · 2026-06-25 | XHigh | 72.5% | LiveBench |
| LiveBench · summarize · 2026-06-25 | XHigh | 72.3% | LiveBench |
| LiveBench · tablejoin · 2026-06-25 | XHigh | 55.9% | LiveBench |
| LiveBench · tablereformat · 2026-06-25 | XHigh | 100.0% | LiveBench |
| LiveBench · theory of mind · 2026-06-25 | XHigh | 84.6% | LiveBench |
| LiveBench · typescript · 2026-06-25 | XHigh | 43.3% | LiveBench |
| LiveBench · typos · 2026-06-25 | XHigh | 88.0% | LiveBench |
| LiveBench · zebra puzzle · 2026-06-25 | XHigh | 100.0% | LiveBench |
| LMArena Agent Arena | Unspecified | 1.0 | LMArena |
| LMArena Agent Arena | XHigh | 4.2 | LMArena |
| MLCR-AA | High | 25.0% | Artificial Analysis |
| MLCR-AA | Medium | 22.2% | Artificial Analysis |
| MLCR-AA | XHigh | 28.9% | Artificial Analysis |
| MMMU-Pro | High | 81.1% | Artificial Analysis |
| MMMU-Pro | Low | 79.0% | Artificial Analysis |
| MMMU-Pro | Medium | 81.2% | Artificial Analysis |
| MMMU-Pro | Non-reasoning | 71.4% | Artificial Analysis |
| MMMU-Pro | XHigh | 79.9% | Artificial Analysis |
| SciCode | High | 56.1% | Artificial Analysis |
| SciCode | Medium | 54.5% | Artificial Analysis |
| SciCode | XHigh | 55.8% | Artificial Analysis |
| SWE Atlas · QnA · codex | XHigh | 45.4% | Scale |
| SWE Atlas · Refactoring · codex | XHigh | 44.8% | Scale |
| SWE Atlas · Test Writing · codex | XHigh | 42.6% | Scale |
| SWE-rebench · 2026-03 · tools | Medium | 60.4% | SWE-rebench |
| SWE-rebench · 2026-03 · tools | XHigh | 63.0% | SWE-rebench |
| SWE-rebench · 2026-04 · tools | Medium | 57.2% | SWE-rebench |
| SWE-rebench · 2026-04 · tools | XHigh | 60.5% | SWE-rebench |
| Terminal-Bench 2.1 | XHigh | 83.2% | Terminal-Bench |
| Terminal-Bench 2.1 | XHigh | 78.0% | Terminal-Bench |
| Terminal-Bench Hard | High | 59.8% | Artificial Analysis |
| Terminal-Bench Hard | Low | 52.3% | Artificial Analysis |
| Terminal-Bench Hard | Medium | 57.6% | Artificial Analysis |
| Terminal-Bench Hard | Non-reasoning | 49.2% | Artificial Analysis |
| Terminal-Bench Hard | XHigh | 60.6% | Artificial Analysis |
| Terminal-Bench v2.1 | High | 79.4% | Artificial Analysis |
| Terminal-Bench v2.1 | Low | 65.5% | Artificial Analysis |
| Terminal-Bench v2.1 | Medium | 80.5% | Artificial Analysis |
| Terminal-Bench v2.1 | Non-reasoning | 61.0% | Artificial Analysis |
| Terminal-Bench v2.1 | XHigh | 84.3% | Artificial Analysis |
| Terminal-Bench v4.0 | High | 9.1% | Artificial Analysis |
| Terminal-Bench v4.0 | Medium | 5.1% | Artificial Analysis |
| Terminal-Bench v4.0 | XHigh | 14.6% | Artificial Analysis |
| Toolathlon-Verified | XHigh | 73.5% | HKUST NLP |
| Vals AI · Code Migration v1 | XHigh | 45.2% | Vals AI |
| Vals AI · Excel Modeling Benchmark v1 | XHigh | 64.5% | Vals AI |
| Vals AI · Finance Agent v2 | XHigh | 51.8% | Vals AI |
| Vals AI · Legal Research Bench v1 | XHigh | 40.4% | Vals AI |
| Vals AI · MMLU-Pro v1 | XHigh | 88.1% | Vals AI |
| Vals AI · Tax Agent Bench v1: Weighted Pass | XHigh | 60.5% | Vals AI |
| Vals AI · Terminal-Bench 2.1 | High | 76.4% | Vals AI |
| Vals AI · Vibe Code Bench v1.1 | XHigh | 69.8% | Vals AI |
| 𝜏²-Bench Telecom | High | 93.0% | Artificial Analysis |
| 𝜏²-Bench Telecom | Low | 83.9% | Artificial Analysis |
| 𝜏²-Bench Telecom | Medium | 91.8% | Artificial Analysis |
| 𝜏²-Bench Telecom | Non-reasoning | 69.3% | Artificial Analysis |
| 𝜏²-Bench Telecom | XHigh | 93.9% | Artificial Analysis |
| 𝜏³-Banking | High | 36.7% | Artificial Analysis |
| 𝜏³-Banking | Low | 24.9% | Artificial Analysis |
| 𝜏³-Banking | Medium | 29.9% | Artificial Analysis |
| 𝜏³-Banking | Non-reasoning | 14.8% | Artificial Analysis |
| 𝜏³-Banking | XHigh | 39.0% | Artificial Analysis |
Benchmark scores & sources
Original results, evaluation harnesses, and evidence behind this model.
| Benchmark | Bucket | Score | Harness | Evidence | Source-recorded date |
|---|---|---|---|---|---|
| SpreadsheetBench v1 · source release snapshot; version not specified | Supporting evidence | 88.1%Reported settings & sourceMiniMax launch figure methodology. SWE internal ClaudeCode 4 runs; Terminal 8CPU16GB2h128k Terminus2; Paper/PostTrain Ralph-loop12h; GDPval internal rubric grader; BrowseComp discard64k; OSWorld361tasks. Comparator provider/leaderboard scores where footnoted. First-party reported result; comparator results retain the source evaluation setup. MiniMax M3 model card benchmark figure · figures/benchmark.jpeg · reviewed 2026-09-06 | Published configuration | lab self-report | Reviewed 2026-09-06 |
| SVG-Bench · source release snapshot; version not specified | Supporting evidence | 58.2%Reported settings & sourceMiniMax launch figure methodology. SWE internal ClaudeCode 4 runs; Terminal 8CPU16GB2h128k Terminus2; Paper/PostTrain Ralph-loop12h; GDPval internal rubric grader; BrowseComp discard64k; OSWorld361tasks. Comparator provider/leaderboard scores where footnoted. First-party reported result; comparator results retain the source evaluation setup. MiniMax M3 model card benchmark figure · figures/benchmark.jpeg · reviewed 2026-09-06 | Published configuration | lab self-report | Reviewed 2026-09-06 |
| SWE-Bench Pro · not specified | Coding | 59.4%Reported settings & sourceLaunch table reported configuration; per-cell reasoning effort unspecified Provider-published result; comparator measurements are not automatically independently reproduced. GPT-5.6: Frontier intelligence that scales with your ambition · Coding table / SWE-Bench Pro / GPT‑5.5 · reviewed 2026-09-06 | Published configurationcontributes to capability | lab self-report | Reviewed 2026-09-06 |
| SWE-bench Pro · source release snapshot; version not specified | Coding | 58.6%Reported settings & sourceMeta Model API xhigh; Gemini high; Opus max; GPT xhigh. Best of self-reported or Meta reproduction unless specified. OSWorld GUI only; Terminal bash only5attempts89tasks6CPU8GB; DeepSWE no internet5attempts. First-party reported result; comparator results retain the source evaluation setup. Muse Spark 1.1 evaluation report Figure44 · Figure44 physical page101; methodology pages101–105 · reviewed 2026-09-06 | Published configurationcontributes to capability | lab self-report | Reviewed 2026-09-06 |
| SWE-bench Pro · source release snapshot; version not specified | Coding | 58.6%Reported settings & sourceMiniMax launch figure methodology. SWE internal ClaudeCode 4 runs; Terminal 8CPU16GB2h128k Terminus2; Paper/PostTrain Ralph-loop12h; GDPval internal rubric grader; BrowseComp discard64k; OSWorld361tasks. Comparator provider/leaderboard scores where footnoted. First-party reported result; comparator results retain the source evaluation setup. MiniMax M3 model card benchmark figure · figures/benchmark.jpeg · reviewed 2026-09-06 | Published configurationcontributes to capability | lab self-report | Reviewed 2026-09-06 |
| SWE-bench Verified · source release snapshot; version not specified | Coding | 82.9%Reported settings & sourceMiniMax launch figure methodology. SWE internal ClaudeCode 4 runs; Terminal 8CPU16GB2h128k Terminus2; Paper/PostTrain Ralph-loop12h; GDPval internal rubric grader; BrowseComp discard64k; OSWorld361tasks. Comparator provider/leaderboard scores where footnoted. First-party reported result; comparator results retain the source evaluation setup. MiniMax M3 model card benchmark figure · figures/benchmark.jpeg · reviewed 2026-09-06 | Published configurationcontributes to capability | lab self-report | Reviewed 2026-09-06 |
| SWE-fficiency · source release snapshot; version not specified | Supporting evidence | 46.6%Reported settings & sourceMiniMax launch figure methodology. SWE internal ClaudeCode 4 runs; Terminal 8CPU16GB2h128k Terminus2; Paper/PostTrain Ralph-loop12h; GDPval internal rubric grader; BrowseComp discard64k; OSWorld361tasks. Comparator provider/leaderboard scores where footnoted. First-party reported result; comparator results retain the source evaluation setup. MiniMax M3 model card benchmark figure · figures/benchmark.jpeg · reviewed 2026-09-06 | Published configuration | lab self-report | Reviewed 2026-09-06 |
| SWE-Marathon · source release snapshot; version not specified | Supporting evidence | 14%Reported settings & sourceTable headers: Kimi K3, GPT-5.6 Sol, Opus 4.8 and GLM-5.2 max; GPT-5.5 xhigh; Fable 5 max with fallbacks. Benchmark-specific footnotes override header settings; cited results are not new Moonshot evaluations. Benchmark-specific tools, harness and budget documented under Evaluation Details. Kimi temp1; single-step top_p.95,agentic top_p1; vision tools=Python,3runs except ZeroBench5runs. First-party reported result. moonshotai/Kimi-K3 · Performance table: SWE-Marathon · reviewed 2026-11-20 | Published configuration | lab self-report | Reviewed 2026-11-20 |
| SWEAtlas-QnA · source release snapshot; version not specified | Supporting evidence | 45.4%Reported settings & sourceMiniMax launch figure methodology. SWE internal ClaudeCode 4 runs; Terminal 8CPU16GB2h128k Terminus2; Paper/PostTrain Ralph-loop12h; GDPval internal rubric grader; BrowseComp discard64k; OSWorld361tasks. Comparator provider/leaderboard scores where footnoted. First-party reported result; comparator results retain the source evaluation setup. MiniMax M3 model card benchmark figure · figures/benchmark.jpeg · reviewed 2026-09-06 | Published configuration | lab self-report | Reviewed 2026-09-06 |
| SWEAtlas-TestWriting · source release snapshot; version not specified | Supporting evidence | 42.6%Reported settings & sourceMiniMax launch figure methodology. SWE internal ClaudeCode 4 runs; Terminal 8CPU16GB2h128k Terminus2; Paper/PostTrain Ralph-loop12h; GDPval internal rubric grader; BrowseComp discard64k; OSWorld361tasks. Comparator provider/leaderboard scores where footnoted. First-party reported result; comparator results retain the source evaluation setup. MiniMax M3 model card benchmark figure · figures/benchmark.jpeg · reviewed 2026-09-06 | Published configuration | lab self-report | Reviewed 2026-09-06 |
| Terminal-Bench 2.1 | Coding | 77.98% | Terminal-Bench 2.1 reported | official board | 2026-04-23 |
| Terminal-Bench 2.1 · 2.1 | Coding | 85.6%Reported settings & sourceLaunch table reported configuration; per-cell reasoning effort unspecified Provider-published result; comparator measurements are not automatically independently reproduced. GPT-5.6: Frontier intelligence that scales with your ambition · Coding table / Terminal-Bench 2.1 / GPT‑5.5 · reviewed 2026-09-06 | Published configuration | lab self-report | Reviewed 2026-09-06 |
| Terminal-Bench 2.1 · 2.1 | Coding | 83.4%Reported settings & sourceTable headers: Kimi K3, GPT-5.6 Sol, Opus 4.8 and GLM-5.2 max; GPT-5.5 xhigh; Fable 5 max with fallbacks. Benchmark-specific footnotes override header settings; cited results are not new Moonshot evaluations. Benchmark-specific tools, harness and budget documented under Evaluation Details. Kimi temp1; single-step top_p.95,agentic top_p1; vision tools=Python,3runs except ZeroBench5runs. Cited result; see benchmark-specific Evaluation Details. Comparison limit: Kimi K3 Evaluation Details cites GLM release blog, Artificial Analysis or OpenAI, per model; retain as published context, not a new Moonshot comparison. moonshotai/Kimi-K3 · Performance table: Terminal-Bench 2.1 · reviewed 2026-11-20 | Published configuration | lab self-report | Reviewed 2026-11-20 |
| Terminal-Bench 2.1 · 2.1 | Coding | 83.4%Reported settings & sourceMeta Model API xhigh; Gemini high; Opus max; GPT xhigh. Best of self-reported or Meta reproduction unless specified. OSWorld GUI only; Terminal bash only5attempts89tasks6CPU8GB; DeepSWE no internet5attempts. First-party reported result; comparator results retain the source evaluation setup. Muse Spark 1.1 evaluation report Figure44 · Figure44 physical page101; methodology pages101–105 · reviewed 2026-09-06 | Published configuration | lab self-report | Reviewed 2026-09-06 |
| Terminal-Bench 2.1 · 2.1 | Coding | 78.2%Reported settings & sourceMiniMax launch figure methodology. SWE internal ClaudeCode 4 runs; Terminal 8CPU16GB2h128k Terminus2; Paper/PostTrain Ralph-loop12h; GDPval internal rubric grader; BrowseComp discard64k; OSWorld361tasks. Comparator provider/leaderboard scores where footnoted. First-party reported result; comparator results retain the source evaluation setup. MiniMax M3 model card benchmark figure · figures/benchmark.jpeg · reviewed 2026-09-06 | Published configuration | lab self-report | Reviewed 2026-09-06 |
| Toolathlon · not specified | Agentic | 55.6%Reported settings & sourceLaunch table reported configuration; per-cell reasoning effort unspecified Provider-published result; comparator measurements are not automatically independently reproduced. GPT-5.6: Frontier intelligence that scales with your ambition · Tool Use table / Toolathlon / GPT‑5.5 · reviewed 2026-09-06 | Published configurationcontributes to capability | lab self-report | Reviewed 2026-09-06 |
| Toolathlon Verified · source release snapshot; version not specified | Agentic | 73.5%Reported settings & sourceMeta Model API xhigh; Gemini high; Opus max; GPT xhigh. Best of self-reported or Meta reproduction unless specified. OSWorld GUI only; Terminal bash only5attempts89tasks6CPU8GB; DeepSWE no internet5attempts. First-party reported result; comparator results retain the source evaluation setup. Muse Spark 1.1 evaluation report Figure44 · Figure44 physical page101; methodology pages101–105 · reviewed 2026-09-06 | Published configurationcontributes to capability | lab self-report | Reviewed 2026-09-06 |
| Toolathlon-Verified · source release snapshot; version not specified | Agentic | 73.5%Reported settings & sourceTable headers: Kimi K3, GPT-5.6 Sol, Opus 4.8 and GLM-5.2 max; GPT-5.5 xhigh; Fable 5 max with fallbacks. Benchmark-specific footnotes override header settings; cited results are not new Moonshot evaluations. Benchmark-specific tools, harness and budget documented under Evaluation Details. Kimi temp1; single-step top_p.95,agentic top_p1; vision tools=Python,3runs except ZeroBench5runs. First-party reported result. moonshotai/Kimi-K3 · Performance table: Toolathlon-Verified · reviewed 2026-11-20 | Published configurationcontributes to capability | lab self-report | Reviewed 2026-11-20 |
| Toolathlon-Verified (official board) | Supporting evidence | 73.5% | Toolathlon-Verified Default agent | official board | 2026-06-30 |
| USAMO 2026 · source release snapshot; version not specified | Supporting evidence | 98.2%Reported settings & sourceMiniMax points out of42; comparator percentages as printed First-party reported result; comparator results retain the source evaluation setup. MiniMax M3 model card benchmark figure · figures/benchmark.jpeg · reviewed 2026-09-06 | Published configuration | lab self-report | Reviewed 2026-09-06 |
| VIBE-V2 · source release snapshot; version not specified | Supporting evidence | 50.5%Reported settings & sourceMiniMax launch figure methodology. SWE internal ClaudeCode 4 runs; Terminal 8CPU16GB2h128k Terminus2; Paper/PostTrain Ralph-loop12h; GDPval internal rubric grader; BrowseComp discard64k; OSWorld361tasks. Comparator provider/leaderboard scores where footnoted. First-party reported result; comparator results retain the source evaluation setup. MiniMax M3 model card benchmark figure · figures/benchmark.jpeg · reviewed 2026-09-06 | Published configuration | lab self-report | Reviewed 2026-09-06 |
| Video-MME (w. sub) · source release snapshot; version not specified | Supporting evidence | 89.3%Reported settings & sourceTable headers: Kimi K3, GPT-5.6 Sol, Opus 4.8 and GLM-5.2 max; GPT-5.5 xhigh; Fable 5 max with fallbacks. Benchmark-specific footnotes override header settings; cited results are not new Moonshot evaluations. Benchmark-specific tools, harness and budget documented under Evaluation Details. Kimi temp1; single-step top_p.95,agentic top_p1; vision tools=Python,3runs except ZeroBench5runs. First-party reported result. moonshotai/Kimi-K3 · Performance table: Video-MME (w. sub) · reviewed 2026-11-20 | Published configuration | lab self-report | Reviewed 2026-11-20 |
| VideoMME with subtitles · source release snapshot; version not specified | Supporting evidence | 89.4%Reported settings & sourceMiniMax launch figure methodology. SWE internal ClaudeCode 4 runs; Terminal 8CPU16GB2h128k Terminus2; Paper/PostTrain Ralph-loop12h; GDPval internal rubric grader; BrowseComp discard64k; OSWorld361tasks. Comparator provider/leaderboard scores where footnoted. First-party reported result; comparator results retain the source evaluation setup. MiniMax M3 model card benchmark figure · figures/benchmark.jpeg · reviewed 2026-09-06 | Published configuration | lab self-report | Reviewed 2026-09-06 |
| VideoMMMU · source release snapshot; version not specified | Supporting evidence | 86.4%Reported settings & sourceMiniMax launch figure methodology. SWE internal ClaudeCode 4 runs; Terminal 8CPU16GB2h128k Terminus2; Paper/PostTrain Ralph-loop12h; GDPval internal rubric grader; BrowseComp discard64k; OSWorld361tasks. Comparator provider/leaderboard scores where footnoted. First-party reported result; comparator results retain the source evaluation setup. MiniMax M3 model card benchmark figure · figures/benchmark.jpeg · reviewed 2026-09-06 | Published configuration | lab self-report | Reviewed 2026-09-06 |
| WebArena Verified · source release snapshot; version not specified | Agentic | 67%Reported settings & sourceMeta Model API xhigh; Gemini high; Opus max; GPT xhigh. Best of self-reported or Meta reproduction unless specified. OSWorld GUI only; Terminal bash only5attempts89tasks6CPU8GB; DeepSWE no internet5attempts. First-party reported result; comparator results retain the source evaluation setup. Muse Spark 1.1 evaluation report Figure44 · Figure44 physical page101; methodology pages101–105 · reviewed 2026-09-06 | Published configurationcontributes to capability | lab self-report | Reviewed 2026-09-06 |