# Claude Opus 4.5High > Source: Benchize (https://benchize.com) — https://benchize.com/models/claude-opus-4-5-high. When citing or reusing this content, credit "Benchize" and link to https://benchize.com/models/claude-opus-4-5-high. Anthropic Canonical URL: https://benchize.com/models/claude-opus-4-5-high Last updated: Thu, 22 Oct 2026 06:30:09 GMT Reproduction and quotation must name Benchize, link to the canonical HTML URL, and retain individual source citations. Anthropic ## Benchmarks | Benchmark | Score | ± | Rank | Cost | Source | | | --- | --- | --- | --- | --- | --- | --- | | [MCP Atlas](https://benchize.com/benchmarks/scale%3Amcp_atlas.md) | 69.8source score | CI 66.9–72.7 | 13 / 22 | — | [scale](https://labs.scale.com/leaderboard/mcp_atlas) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 69.8 \| source_score \| reported \| Higher is better \| ### Configuration Reasoning effort high Model configuration claude-opus-4-5 (high) sourceModelVersion Not reported thinkingEnabled true ### Reported statistics Uncertainty Confidence interval: 66.9–72.7 contaminationWarning Not reported rank 22 reportedConfidenceHalfWidth 2.9 ### Source [Original source](https://labs.scale.com/leaderboard/mcp_atlas) Source scale Original model label claude-opus-4-5 (high) Observed 2026-10-22T03:30:07.875Z Evidence SHA-256 `7b0bc178e35a1688340e9efca5795a8f54d07e56a98f9cacd5e4471e0da266e9` Source locator `/rows/21` | | | | | | | | [WeirdML v3](https://benchize.com/benchmarks/weirdml%3Av3.md) | 0.07source score | Range 0.05–0.08 | 19 / 25 | — | [weirdml · b0120fe](https://htihle.github.io/assets/data/weirdml_v3.json) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 0.06514876515151515 \| source_score \| reported \| Higher is better \| \| `mean_final_best` \| 0.07125567424242425 \| source_score \| reported \| Higher is better \| ### Configuration budget {"kind":"equivalent_tokens","limit":50000000,"start":500000} harness claude_code Reasoning effort high sourceConfiguration opus-4.5 thinkingEnabled true ### Reported statistics Uncertainty Reported interval: 0.05–0.08 interval [0.049867746775283606,0.08029579747854858] intervalMethod {"assumptions":"Independent fresh runs; approximately normal scores; shared variance prior. Pool fitted within analysis mode, including incomplete models; pool estimation uncertainty is not propagated.","confidence":0.95,"draws":10000,"fixed_tasks":true,"method":"partially_pooled_variance","pool":{"degrees_of_freedom":1533,"fallback":false,"prior_df":4,"variance":0.0026383113392869583},"seed":20260915,"unit":"run"} runCount 42 taskCount 11 ### Source [Original source](https://htihle.github.io/assets/data/weirdml_v3.json) Source weirdml Original model label Claude Opus 4.5 (high) Source version b0120fe Observed 2026-10-22T04:50:05.075Z Evidence SHA-256 `82c3a8de423b23ca0af9761429299a2f21edb0b819e947fdc863f8b5ea4a49b8` Source locator `/rows/18` | | | | | | | | [τ-bench banking knowledge](https://benchize.com/benchmarks/sierra%3Abanking_knowledge.md) | 24.74% | — | 15 / 19 | — | [sierra](https://sierra-tau-bench-public.s3.us-west-2.amazonaws.com/submissions/claude-opus-4-5_sierra_2026-02-26/submission.json) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `pass_1` \| 24.74 \| percent \| consistent_success_across_k_trials \| Higher is better \| \| `pass_2` \| 17.18 \| percent \| consistent_success_across_k_trials \| Higher is better \| \| `pass_3` \| 13.66 \| percent \| consistent_success_across_k_trials \| Higher is better \| \| `pass_4` \| 11.34 \| percent \| consistent_success_across_k_trials \| Higher is better \| ### Configuration benchmarkVersion 1.0.1 harness tau-bench methodology {"evaluation_date":"2026-05-05","notes":"Evaluated using AllTools retrieval (BM25 + dense OpenAI text-embedding-3-large + sandboxed shell). User simulator: gpt-5.2 with reasoning_effort: low. 4 trials. Seed: 300. Banking_knowledge domain only — other domains intentionally excluded from this comparison; the AllTools setting standardizes retrieval across models. banking_knowledge re-graded under tau2-bench v1.0.1 grading fixes on 2026-07-15 (extra-read logging #329, numeric-argument normalization #397, T077-T086 gold trajectories #402); scores prior to v1.0.1 are not comparable.","tau2_bench_version":"1.0.1","user_simulator":"gpt-5.2","verification":{"details":"Verified evaluation with full trajectory data available. Standard tau-bench scaffold.","modified_prompts":false,"omitted_questions":false}} modality text Reasoning effort high retrieval alltools sourceSubmission claude-opus-4-5_sierra_2026-02-26 ### Source [Original source](https://sierra-tau-bench-public.s3.us-west-2.amazonaws.com/submissions/claude-opus-4-5_sierra_2026-02-26/submission.json) Source sierra Original model label Claude Opus 4.5 Evaluated 2026-02-26 Observed 2026-10-21T16:54:46.911Z Evidence SHA-256 `9c3b8b400e21d370cea27caec3643bfe6fd8385be9ed5135b585d0b976b6e34f` Source locator `/results/banking_knowledge` | | | | | | | | [τ-bench retail](https://benchize.com/benchmarks/sierra%3Aretail.md) | 79.61% | — | 3 / 8 | $0.39 | [sierra](https://sierra-tau-bench-public.s3.us-west-2.amazonaws.com/submissions/claude-opus-4-5_sierra_2026-02-26/submission.json) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `pass_1` \| 79.61 \| percent \| consistent_success_across_k_trials \| Higher is better \| \| `pass_2` \| 67.4 \| percent \| consistent_success_across_k_trials \| Higher is better \| \| `pass_3` \| 58.77 \| percent \| consistent_success_across_k_trials \| Higher is better \| \| `pass_4` \| 51.75 \| percent \| consistent_success_across_k_trials \| Higher is better \| \| `cost` \| 0.38695 \| USD \| source_reported \| Lower is better \| ### Configuration benchmarkVersion 1.0.1 harness tau-bench methodology {"evaluation_date":"2026-05-05","notes":"Evaluated using AllTools retrieval (BM25 + dense OpenAI text-embedding-3-large + sandboxed shell). User simulator: gpt-5.2 with reasoning_effort: low. 4 trials. Seed: 300. Banking_knowledge domain only — other domains intentionally excluded from this comparison; the AllTools setting standardizes retrieval across models. banking_knowledge re-graded under tau2-bench v1.0.1 grading fixes on 2026-07-15 (extra-read logging #329, numeric-argument normalization #397, T077-T086 gold trajectories #402); scores prior to v1.0.1 are not comparable.","tau2_bench_version":"1.0.1","user_simulator":"gpt-5.2","verification":{"details":"Verified evaluation with full trajectory data available. Standard tau-bench scaffold.","modified_prompts":false,"omitted_questions":false}} modality text Reasoning effort high retrieval Not reported sourceSubmission claude-opus-4-5_sierra_2026-02-26 ### Source [Original source](https://sierra-tau-bench-public.s3.us-west-2.amazonaws.com/submissions/claude-opus-4-5_sierra_2026-02-26/submission.json) Source sierra Original model label Claude Opus 4.5 Evaluated 2026-02-26 Observed 2026-10-21T16:54:46.911Z Evidence SHA-256 `9c3b8b400e21d370cea27caec3643bfe6fd8385be9ed5135b585d0b976b6e34f` Source locator `/results/retail` | | | | | | | | [τ-bench airline](https://benchize.com/benchmarks/sierra%3Aairline.md) | 84% | — | 1 / 8 | $0.4 | [sierra](https://sierra-tau-bench-public.s3.us-west-2.amazonaws.com/submissions/claude-opus-4-5_sierra_2026-02-26/submission.json) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `pass_1` \| 84 \| percent \| consistent_success_across_k_trials \| Higher is better \| \| `pass_2` \| 77.67 \| percent \| consistent_success_across_k_trials \| Higher is better \| \| `pass_3` \| 73.5 \| percent \| consistent_success_across_k_trials \| Higher is better \| \| `pass_4` \| 70 \| percent \| consistent_success_across_k_trials \| Higher is better \| \| `cost` \| 0.39919 \| USD \| source_reported \| Lower is better \| ### Configuration benchmarkVersion 1.0.1 harness tau-bench methodology {"evaluation_date":"2026-05-05","notes":"Evaluated using AllTools retrieval (BM25 + dense OpenAI text-embedding-3-large + sandboxed shell). User simulator: gpt-5.2 with reasoning_effort: low. 4 trials. Seed: 300. Banking_knowledge domain only — other domains intentionally excluded from this comparison; the AllTools setting standardizes retrieval across models. banking_knowledge re-graded under tau2-bench v1.0.1 grading fixes on 2026-07-15 (extra-read logging #329, numeric-argument normalization #397, T077-T086 gold trajectories #402); scores prior to v1.0.1 are not comparable.","tau2_bench_version":"1.0.1","user_simulator":"gpt-5.2","verification":{"details":"Verified evaluation with full trajectory data available. Standard tau-bench scaffold.","modified_prompts":false,"omitted_questions":false}} modality text Reasoning effort high retrieval Not reported sourceSubmission claude-opus-4-5_sierra_2026-02-26 ### Source [Original source](https://sierra-tau-bench-public.s3.us-west-2.amazonaws.com/submissions/claude-opus-4-5_sierra_2026-02-26/submission.json) Source sierra Original model label Claude Opus 4.5 Evaluated 2026-02-26 Observed 2026-10-21T16:54:46.911Z Evidence SHA-256 `9c3b8b400e21d370cea27caec3643bfe6fd8385be9ed5135b585d0b976b6e34f` Source locator `/results/airline` | | | | | | | | [BullshitBench v2](https://benchize.com/benchmarks/bullshitbench%3Av2.md) | 90% | — | 5 / 191 | — | [bullshitbench · v2](https://api.github.com/repos/petergpt/bullshit-benchmark/contents/data/v2/latest/leaderboard.csv?ref=3c655234d15967738f95da696e3f938d4862dcf7) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `green_rate` \| 0.9 \| score_fraction \| all_attempts \| Higher is better \| \| `red_rate` \| 0.02 \| score_fraction \| all_attempts \| Lower is better \| \| `refusal_rate` \| 0.0 \| score_fraction \| all_attempts \| Lower is better \| \| `avg_score` \| 1.8433 \| score \| judge_grade_0_to_2 \| Higher is better \| ### Configuration Reasoning effort high suite v2 thinkingEnabled true ### Reported statistics acceptedCount 2 answeredCount 100 clearCount 90 errorCount 0 partialCount 8 rank 6 refusalCount 0 sampleSize 100 ### Source [Original source](https://api.github.com/repos/petergpt/bullshit-benchmark/contents/data/v2/latest/leaderboard.csv?ref=3c655234d15967738f95da696e3f938d4862dcf7) Source bullshitbench Original model label anthropic/claude-opus-4.5 Source version v2 Observed 2026-10-22T03:10:12.603Z Evidence SHA-256 `98b485e08764d1c97144bc4f7b28ced1a96d50850947c775870dfda0e5d749b5` Source locator `/rows/5` | | | | | | | | [τ-bench telecom](https://benchize.com/benchmarks/sierra%3Atelecom.md) | 92.32% | — | 2 / 8 | $0.72 | [sierra](https://sierra-tau-bench-public.s3.us-west-2.amazonaws.com/submissions/claude-opus-4-5_sierra_2026-02-26/submission.json) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `pass_1` \| 92.32 \| percent \| consistent_success_across_k_trials \| Higher is better \| \| `pass_2` \| 86.11 \| percent \| consistent_success_across_k_trials \| Higher is better \| \| `pass_3` \| 81.36 \| percent \| consistent_success_across_k_trials \| Higher is better \| \| `pass_4` \| 78.07 \| percent \| consistent_success_across_k_trials \| Higher is better \| \| `cost` \| 0.71992 \| USD \| source_reported \| Lower is better \| ### Configuration benchmarkVersion 1.0.1 harness tau-bench methodology {"evaluation_date":"2026-05-05","notes":"Evaluated using AllTools retrieval (BM25 + dense OpenAI text-embedding-3-large + sandboxed shell). User simulator: gpt-5.2 with reasoning_effort: low. 4 trials. Seed: 300. Banking_knowledge domain only — other domains intentionally excluded from this comparison; the AllTools setting standardizes retrieval across models. banking_knowledge re-graded under tau2-bench v1.0.1 grading fixes on 2026-07-15 (extra-read logging #329, numeric-argument normalization #397, T077-T086 gold trajectories #402); scores prior to v1.0.1 are not comparable.","tau2_bench_version":"1.0.1","user_simulator":"gpt-5.2","verification":{"details":"Verified evaluation with full trajectory data available. Standard tau-bench scaffold.","modified_prompts":false,"omitted_questions":false}} modality text Reasoning effort high retrieval Not reported sourceSubmission claude-opus-4-5_sierra_2026-02-26 ### Source [Original source](https://sierra-tau-bench-public.s3.us-west-2.amazonaws.com/submissions/claude-opus-4-5_sierra_2026-02-26/submission.json) Source sierra Original model label Claude Opus 4.5 Evaluated 2026-02-26 Observed 2026-10-21T16:54:46.911Z Evidence SHA-256 `9c3b8b400e21d370cea27caec3643bfe6fd8385be9ed5135b585d0b976b6e34f` Source locator `/results/telecom` | | | | | | | --- Cite as: Claude Opus 4.5High — Benchize, https://benchize.com/models/claude-opus-4-5-high (retrieved 2026-10-22)