RankingClaude Sonnet 4.5

Claude Sonnet 4.5

Data updated 22 Oct 2026

30 published benchmark measures · 14 benchmark families contribute across 6 task areas. 3 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

Adjusted comparison score · 0–100

Filled points are supported; hollow points are preliminary. Lines stop at unknown capabilities. Exact values and sources follow below.

Agentic4 families · 1 with independent evidence · Supported1.1

4 core families; 19 direct opponents across 8 labs.

Estimated against the same complete panel: gpt-5.6-sol claude-opus-4-8 gemini-3.1-pro . Indirect comparisons assume performance can be summarized by one strength within this capability.

Refit without one family: 0.3–10.9; 0/13 scenarios unsupported. Without one publisher: 0.2–14.1; 0/15 unsupported. Smoothing check: 0.1–7.3; 0/3 unsupported. These are sensitivity checks, not confidence intervals.

Audit: family and publisher holdouts improved on neutral predictions within evaluated coverage; source sensitivity still requires inspection. Validation audit.

Matched family win shares below describe source evidence; they are not averaged to produce the adjusted score.

OSWorld: 20.0 observed win share
os-world.github.io · Source 1
Bfcl V4: 100.0 observed win share
Amazon · Source 1
MCP-Atlas: 75.0 observed win share
Amazon · Source 1
Browsecomp: 0.0 observed win share
Mistral · Source 1

Hard reasoning4 families · 2 with independent evidence · Supported9.2

4 core families; 71 direct opponents across 12 labs.

Estimated against the same complete panel: gpt-5.6-sol claude-opus-4-8 gemini-3.1-pro . Indirect comparisons assume performance can be summarized by one strength within this capability.

Refit without one family: 6.8–14.8; 0/6 scenarios unsupported. Without one publisher: 6.5–10.0; 0/16 unsupported. Smoothing check: 3.0–21.7; 0/3 unsupported. These are sensitivity checks, not confidence intervals.

Audit: family and publisher holdouts improved on neutral predictions within evaluated coverage. Validation audit.

Matched family win shares below describe source evidence; they are not averaged to produce the adjusted score.

ARC-AGI: 18.6 observed win share
arcprize.org · Source 1
Humanity’s Last Exam: 37.5 observed win share
scale.com · Source 1
Aime: 47.7 observed win share
Amazon, Mistral · Source 1 Source 2
GPQA: 72.7 observed win share
Amazon · Source 1

Coding2 families · 1 with independent evidence · Supported1.8

2 core families; 28 direct opponents across 10 labs.

Estimated against the same complete panel: gpt-5.6-sol claude-opus-4-8 gemini-3.1-pro . Indirect comparisons assume performance can be summarized by one strength within this capability.

Refit without one family: 0.7–8.7; 2/6 scenarios unsupported. Without one publisher: 1.0–7.9; 1/19 unsupported. Smoothing check: 0.7–6.0; 0/3 unsupported. These are sensitivity checks, not confidence intervals.

Audit: family and publisher holdouts improved on neutral predictions within evaluated coverage. Validation audit.

Matched family win shares below describe source evidence; they are not averaged to produce the adjusted score.

SWE-bench Verified: 65.7 observed win share
Amazon, Mistral, swebench.com · Source 1 Source 2 Source 3 Source 4
LiveCodeBench: 44.4 observed win share
Amazon · Source 1

Human pref1 families · 1 with independent evidence · PreliminaryUnknown

1 core families; 41 direct opponents across 13 labs. No connected comparison to the complete reference panel

Estimated against the same complete panel: gpt-5.6-sol claude-opus-4-8 gemini-3.1-pro . Indirect comparisons assume performance can be summarized by one strength within this capability.

Refit without one family: No supported estimate; 1/1 scenarios unsupported. Without one publisher: No supported estimate; 1/1 unsupported. Smoothing check: No supported estimate; 3/3 unsupported. These are sensitivity checks, not confidence intervals.

Audit: cross-family and cross-publisher predictions could not be evaluated from the available comparison network. Validation audit.

Matched family win shares below describe source evidence; they are not averaged to produce the adjusted score.

Human preference (Arena): 87.8 observed win share
lmarena.ai · Source 1

Knowledge1 families · 0 with independent evidence · PreliminaryUnknown

1 core families; 11 direct opponents across 4 labs. No connected comparison to the complete reference panel

Estimated against the same complete panel: gpt-5.6-sol claude-opus-4-8 gemini-3.1-pro . Indirect comparisons assume performance can be summarized by one strength within this capability.

Refit without one family: No supported estimate; 4/4 scenarios unsupported. Without one publisher: No supported estimate; 5/5 unsupported. Smoothing check: No supported estimate; 3/3 unsupported. These are sensitivity checks, not confidence intervals.

Audit: almost no publisher-held-out comparisons remained estimable. Cross-source validity is unresolved. Validation audit.

Matched family win shares below describe source evidence; they are not averaged to produce the adjusted score.

MMLU-Pro: 100.0 observed win share
Amazon · Source 1

Multimodal2 families · 0 with independent evidence · PreliminaryUnknown

2 core families; 11 direct opponents across 4 labs. No connected comparison to the complete reference panel

Estimated against the same complete panel: gpt-5.6-sol claude-opus-4-8 gemini-3.1-pro . Indirect comparisons assume performance can be summarized by one strength within this capability.

Refit without one family: No supported estimate; 8/8 scenarios unsupported. Without one publisher: No supported estimate; 8/8 unsupported. Smoothing check: No supported estimate; 3/3 unsupported. These are sensitivity checks, not confidence intervals.

Audit: limited publisher-held-out coverage. Cross-source validity is unresolved. Validation audit.

Matched family win shares below describe source evidence; they are not averaged to produce the adjusted score.

Ocrbench V2 Average Accuracy: 9.1 observed win share
Amazon · Source 1
Mmmu Pro: 68.2 observed win share
Amazon · Source 1

Long contextNo comparable evidenceUnknown

0 core families; 0 direct opponents across 0 labs. No admitted core comparisons

Estimated against the same complete panel: gpt-5.6-sol claude-opus-4-8 gemini-3.1-pro . Indirect comparisons assume performance can be summarized by one strength within this capability.

Refit without one family: No supported estimate; 5/5 scenarios unsupported. Without one publisher: No supported estimate; 5/5 unsupported. Smoothing check: No supported estimate; 3/3 unsupported. These are sensitivity checks, not confidence intervals.

Audit: family and publisher holdouts underperformed neutral predictions. Treat this as an exploratory estimate. Validation audit.

Matched family win shares below describe source evidence; they are not averaged to produce the adjusted score.

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 · Not specified

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.

Not specified: 33 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 evidence

Score contributions and missing evidence

14 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: Anthropic
Catalog status: active
Availability: Public provider catalog; account and region restrictions may apply
Family: Claude Sonnet
Released: —
Context: —
License: proprietary
Model card: https://platform.claude.com/docs/en/about-claude/model-deprecations
Default Capability family coverage: /badge/claude-sonnet-4-5-20250929.svg

Benchmark scores & sources

Original results, evaluation harnesses, and evidence behind this model.

Benchmark	Bucket	Score	Harness	Evidence	Source-recorded date
tau2 Airline Verified · source release snapshot; version not specified	Supporting evidence	66.8% Reported settings & source Nova2 launch evaluation; benchmark-specific settings in report. tau2 Verified avg@3; telecom cited AA because user model sensitivity; comparators mix own runs and cited provider scores. First-party reported result; comparator results retain the source evaluation setup. Amazon Nova 2 technical report · Table2 · reviewed 2026-09-06	Published configuration Effort: Not specified	lab self-report Supporting evidence outside the reviewed capability core	Reviewed 2026-09-06
tau2 Retail Verified · source release snapshot; version not specified	Supporting evidence	77.2% Reported settings & source Nova2 launch evaluation; benchmark-specific settings in report. tau2 Verified avg@3; telecom cited AA because user model sensitivity; comparators mix own runs and cited provider scores. First-party reported result; comparator results retain the source evaluation setup. Amazon Nova 2 technical report · Table2 · reviewed 2026-09-06	Published configuration Effort: Not specified	lab self-report Supporting evidence outside the reviewed capability core	Reviewed 2026-09-06
tau2 Telecom · source release snapshot; version not specified	Supporting evidence	78.1% Reported settings & source Nova2 launch evaluation; benchmark-specific settings in report. tau2 Verified avg@3; telecom cited AA because user model sensitivity; comparators mix own runs and cited provider scores. First-party reported result; comparator results retain the source evaluation setup. Amazon Nova 2 technical report · Table2 · reviewed 2026-09-06	Published configuration Effort: Not specified	lab self-report Supporting evidence outside the reviewed capability core	Reviewed 2026-09-06
tau3 Airline · source release snapshot; version not specified	Supporting evidence	72% Reported settings & source Maximum reasoning; tau3 4 trials, GPT5.2 low simulator except Sierra reported comparisons; BrowseComp discard-all context at100k. First-party reported result; comparator results retain the source evaluation setup. Mistral Medium 3.5 model card performance charts · images/image4.png · reviewed 2026-09-06	Published configuration Effort: Not specified	lab self-report Supporting evidence outside the reviewed capability core	Reviewed 2026-09-06
tau3 Banking · source release snapshot; version not specified	Supporting evidence	22.4% Reported settings & source Maximum reasoning; tau3 4 trials, GPT5.2 low simulator except Sierra reported comparisons; BrowseComp discard-all context at100k. First-party reported result; comparator results retain the source evaluation setup. Mistral Medium 3.5 model card performance charts · images/image4.png · reviewed 2026-09-06	Published configuration Effort: Not specified	lab self-report Supporting evidence outside the reviewed capability core	Reviewed 2026-09-06
tau3 Retail · source release snapshot; version not specified	Supporting evidence	72.4% Reported settings & source Maximum reasoning; tau3 4 trials, GPT5.2 low simulator except Sierra reported comparisons; BrowseComp discard-all context at100k. First-party reported result; comparator results retain the source evaluation setup. Mistral Medium 3.5 model card performance charts · images/image4.png · reviewed 2026-09-06	Published configuration Effort: Not specified	lab self-report Supporting evidence outside the reviewed capability core	Reviewed 2026-09-06
tau3 Telecom · source release snapshot; version not specified	Supporting evidence	84.9% Reported settings & source Maximum reasoning; tau3 4 trials, GPT5.2 low simulator except Sierra reported comparisons; BrowseComp discard-all context at100k. First-party reported result; comparator results retain the source evaluation setup. Mistral Medium 3.5 model card performance charts · images/image4.png · reviewed 2026-09-06	Published configuration Effort: Not specified	lab self-report Supporting evidence outside the reviewed capability core	Reviewed 2026-09-06
Terminal-Bench 1.0 · 1.0	Coding	51% Reported settings & source Nova2 internal agentic scaffold; scaled inference explicitly separate. Comparator setups from cited provider report. First-party reported result; comparator results retain the source evaluation setup. Amazon Nova 2 technical report · Table4 · reviewed 2026-09-06	Published configuration Effort: Not specified	lab self-report Terminal-Bench coding slot admits version 4.0 only. Earlier versions remain stored and are not mixed into that slot.	Reviewed 2026-09-06
DeepSWE v1.1	Agentic	—	—	—	—
GDPval-AA	Agentic	—	—	—	—
GPQA Diamond	Hard reasoning	—	—	—	—
LiveCodeBench	Coding	—	—	—	—
MMLU-Pro	Knowledge	—	—	—	—
SWE-bench Pro	Agentic	—	—	—	—
Terminal-Bench 2.1	Agentic	—	—	—	—