BlogLLM Benchmark Scores as Published, Side by Side
Claude Sonnet 4: Benchmark Scores as Published
Every result the catalog holds for Claude Sonnet 4: 37 tests in 10 areas, from 7 pages. 13 printed by Anthropic itself, the rest by the company named beside each number. Each number is as printed; nothing is ranked or rescored.
Developer AnthropicReleased 2025-05-22Data version 2026103001, as of 2026-10-30
Knowledge Work
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| Vals Finance Agent (version not stated) | 44.5% | Anthropic, Claude Sonnet 4.5checked 2026-10-15 | Effort: extended thinking (up to 64K) |
Agentic Coding
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| SWE-bench Verified | 80.2% | Anthropic, Claude Sonnet 4.5checked 2026-10-15; printed on 4 more pages | Effort: thinking budget 200K; Attempt budget: parallel test-time compute: multiple samples, the best selected by an internal scoring model |
| 72.7% | Anthropic, Claude Sonnet 4.5checked 2026-10-15 | Effort: thinking budget 200K | |
| SWE-bench Multilingual | 51% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | Effort: no extended thinking; Harness: agentic, single attempt |
| Terminal-Bench (version not stated) | 36.4% | Anthropic, Claude Sonnet 4.5checked 2026-10-15; printed on 4 more pages | - |
| SWE-bench Verified (agentless, single patch without test) | 50.2% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | Effort: no extended thinking; Harness: agentless |
| SWE-bench Verified (multiple attempts) | 80.2% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | Effort: no extended thinking; Harness: agentic, multiple attempts |
Science and Math
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| GPQA Diamond | 76.1% | Anthropic, Claude Sonnet 4.5checked 2026-10-15; printed on 4 more pages | - |
| AIME 2025 | 70.5% | Anthropic, Claude Sonnet 4.5checked 2026-10-15; printed on 4 more pages | Tools: none |
| HMMT February 2025 | 15.9% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | Effort: no extended thinking |
| AIME 2024 | 43.4% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | Effort: no extended thinking |
| MATH-500 | 94% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | Effort: no extended thinking |
| CNMO 2024 | 60.4% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | Effort: no extended thinking |
| PolyMath (en) | 52.8% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | Effort: no extended thinking |
Computer Use
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| OSWorld-Verified | 42.2% | Anthropic, Claude Sonnet 4.5checked 2026-10-15; printed on 2 more pages | - |
Multimodal Understanding
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| MMMU (validation) | 74.4% | Anthropic, Claude Sonnet 4.5checked 2026-10-15; printed on 3 more pages | - |
General Knowledge and Reasoning
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| Humanity's Last Exam (no tools) | 5.8% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | Effort: no extended thinking |
| MMLU-Pro | 83.7% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | Effort: no extended thinking |
| MMLU-Redux | 93.6% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | Effort: no extended thinking |
| SuperGPQA | 55.7% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | Effort: no extended thinking |
| MMLU | 91.5% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | Effort: no extended thinking |
| SimpleQA | 15.9% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | Effort: no extended thinking |
| AutoLogi | 89.8% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | Effort: no extended thinking |
| LiveBench (2024-11-25) | 74.8% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | Effort: no extended thinking |
| ZebraLogic | 79.7% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | Effort: no extended thinking |
Code Generation
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| LiveCodeBench v6 | 48.5% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | Effort: no extended thinking |
| Aider polyglot | 56.4% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | Effort: no extended thinking |
| MultiPL-E | 88.6% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | Effort: no extended thinking |
| OJBench (language not stated) | 15.3% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | Effort: no extended thinking |
Tool Use and Workflows
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| τ²-bench Telecom | 49.6% | Anthropic, Claude Sonnet 4.5checked 2026-10-15; printed on 2 more pages | Effort: extended thinking (up to 64K) |
| τ²-bench Retail | 83.8% | Anthropic, Claude Sonnet 4.5checked 2026-10-15; printed on 2 more pages | Effort: extended thinking (up to 64K) |
| τ²-bench Airline | 63% | Anthropic, Claude Sonnet 4.5checked 2026-10-15; printed on 2 more pages | Effort: extended thinking (up to 64K) |
| τ-bench Airline | 60% | Anthropic, Claude Opus 4 and Claude Sonnet 4checked 2026-10-15; printed on 1 more page | Effort: extended thinking, up to 64K |
| τ-bench Retail | 80.5% | Anthropic, Claude Opus 4 and Claude Sonnet 4checked 2026-10-15; printed on 1 more page | Effort: extended thinking, up to 64K |
| AceBench | 76.2% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | Effort: no extended thinking |
Instruction Following
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| MultiChallenge | 46.8% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | Effort: no extended thinking |
| IFEval (prompt-level strict) | 87.6% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | Effort: no extended thinking |
Multilingual
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| MMMLU | 86.5% | Anthropic, Claude Sonnet 4.5checked 2026-10-15; printed on 3 more pages | Effort: extended thinking (up to 128K) |
Price
API list prices in US dollars per one million tokens, as printed on the developer's pricing page on the day checked.
| Price | Value | Printed by | Setup as printed |
|---|---|---|---|
| Input | $3 | Anthropic, Anthropic API pricingchecked 2026-10-15 | - |
| Cached input | $0.3 | Anthropic, Anthropic API pricingchecked 2026-10-15 | - |
| Cache write | $3.75 | Anthropic, Anthropic API pricingchecked 2026-10-15 | Cache: 5 minute cache |
| Cache write, 1 hour | $6 | Anthropic, Anthropic API pricingchecked 2026-10-15 | - |
| Output | $15 | Anthropic, Anthropic API pricingchecked 2026-10-15 | - |
| Input, batch | $1.5 | Anthropic, Anthropic API pricingchecked 2026-10-15 | - |
| Output, batch | $7.5 | Anthropic, Anthropic API pricingchecked 2026-10-15 | - |
More from Anthropic
- Claude Haiku 5.5
- Claude Sonnet 5.5
- Claude Opus 5.5
- Claude Fable 5.1
- Claude Mythos 5.1
- Claude Opus 5
- Claude Sonnet 5
- Claude Fable 5
- Claude Mythos 5
- Claude Opus 4.8
- Claude Opus 4.7
- Claude Mythos Preview
- Claude Sonnet 4.6
- Claude Opus 4.6
- Claude Opus 4.5
- Claude Haiku 4.5
- Claude Sonnet 4.5
- Claude Opus 4.1
- Claude Opus 4
- Claude 3.7 Sonnet
- Claude 3.5 Sonnet
Every number here is read from the same catalog file as the benchmark table, data version 2026103001. Where one page printed the model under several setups, each value is listed. The catalog is public in armanobosyan/sugra-llm-benchmarks; a wrong number can be reported from the table, with Report an error under the value, or as an issue in that repository.