Model comparison
Claude Haiku 4.5 vs GPT-6 Astra
GPT-6 Astra is the stronger model overall, scoring 70.8 to 39.5 on the Noometry Index. Claude Haiku 4.5 costs 10× less per token, which makes it the better buy when GPT-6 Astra's lead doesn't matter for your workload.
Last verified . 37 shared benchmarks.
Summary
- They share 37 benchmarks with published results for both. Claude Haiku 4.5 scores higher in 0 categories and GPT-6 Astra in 10 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where GPT-6 Astra leads 85.1 to 15.1.
- The biggest single-benchmark swing is ARC-AGI-2: 4% for Claude Haiku 4.5 and 95% for GPT-6 Astra.
- Claude Haiku 4.5 is cheaper at $1 / $5 per million input/output tokens, against $10 / $50 for GPT-6 Astra.
- GPT-6 Astra accepts more context: 1.05M tokens versus 200K.
Side by side
| Claude Haiku 4.5 | GPT-6 Astra | |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 39.5 | 70.8 |
| Released | 2025-10-15 | 2026-09-03 |
| Weights | Proprietary | Proprietary |
| Context window | 200K | 1.05M |
| Max output | 64K | 128K |
| Input $ / M tokens | $1 | $10 |
| Output $ / M tokens | $5 | $50 |
| Results tracked | 53 | 56 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-6 Astra leads
Claude Haiku 4.5: 44.0 (#78), GPT-6 Astra: 73.7 (#2)
| Benchmark | Claude Haiku 4.5 | GPT-6 Astra |
|---|---|---|
| LMArena WebDev | 1330 | 1786 |
| SciCode | 43.3% | 56.5% |
| WeirdML | 45.4% | 93.6% |
| LMArena Coding | 1453 | 1487 |
| ALE-Bench | 653.48 | 2,951 |
| DeepSWE | — | 74.1% |
| FrontierCode | — | 53.3% |
| SWE-bench Verified (bash only) | 66.6% | — |
| SWE-bench Multilingual | 64.7% | — |
| FrontierSWE | — | 65.5% |
| GSO | — | 79.4% |
| MirrorCode | — | 46.7% |
Agentic & Tool Use GPT-6 Astra leads
Claude Haiku 4.5: 33.6 (#52), GPT-6 Astra: 52.9 (#3)
| Benchmark | Claude Haiku 4.5 | GPT-6 Astra |
|---|---|---|
| BALROG | 31.2% | 68.3% |
| Vending-Bench 2 | 458.89 | 15,515 |
| Terminal-Bench | 35.5% | — |
| APEX-Agents | — | 64.7% |
| Berkeley Function Calling Leaderboard | 68.7% | — |
| Remote Labor Index | — | 20.8% |
| DeepResearch Bench | 45.5% | — |
| ExploitBench | 13.7% | — |
| GDP.pdf | — | 34.2% |
Reasoning GPT-6 Astra leads
Claude Haiku 4.5: 15.1 (#320), GPT-6 Astra: 85.1 (#1)
| Benchmark | Claude Haiku 4.5 | GPT-6 Astra |
|---|---|---|
| ARC-AGI-2 | 4% | 95% |
| NYT Connections (extended) | 14.3% | 98.1% |
| ARC-AGI-1 | 47.7% | 98.5% |
| CritPt | 0% | 31.7% |
| Chess Puzzles | 8% | 72% |
| LMArena Hard Prompts | 1420 | 1462 |
| DTBench | 73.6% | 97.3% |
| LMCA | 30.9% | 64.4% |
| Epoch Capabilities Index | 142.41 | 166.45 |
| EBR-Bench | — | 76.2% |
| Mystery Game Puzzles | — | 84% |
| Bench to the Future 3 | — | 0.14 |
| ForecastBench | 61.4 | — |
Math GPT-6 Astra leads
Claude Haiku 4.5: 44.9 (#78), GPT-6 Astra: 93.5 (#2)
| Benchmark | Claude Haiku 4.5 | GPT-6 Astra |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 66.7% | 100% |
| LMArena Math | 1396 | 1465 |
| FrontierMath (Tiers 1-3) | — | 93.7% |
| FrontierMath Tier 4 | — | 97.6% |
| ProofBench | — | 99% |
| Omni-MATH | 56.1% | — |
| MATH Level 5 | 96.4% | — |
| FrontierMath (Feb 2025 set) | 5.9% | — |
| FrontierMath Erdős | — | 2.9% |
| FrontierMath Tier 4 (v1) | 2.1% | — |
Knowledge GPT-6 Astra leads
Claude Haiku 4.5: 37.7 (#153), GPT-6 Astra: 75.3 (#1)
| Benchmark | Claude Haiku 4.5 | GPT-6 Astra |
|---|---|---|
| GPQA Diamond | 71.2% | 95.8% |
| SimpleQA Verified | 13.2% | 75.6% |
| Vectara Hallucination Rate | 9.8% | 8.7% |
| LMArena Expert | 1442 | 1483 |
| Humanity's Last Exam | — | 54.8% |
| MMLU-Pro | 77.7% | — |
| GPQA (HELM) | 60.5% | — |
Multimodal GPT-6 Astra leads
Claude Haiku 4.5: 26.8 (#118), GPT-6 Astra: 55.0 (#3)
| Benchmark | Claude Haiku 4.5 | GPT-6 Astra |
|---|---|---|
| Blueprint-Bench 2 | 0% | 49.7% |
| LMArena Document | 1420 | 1468 |
| LMArena Vision | — | 1281 |
| Furniture Assembly | — | 80% |
Multilingual GPT-6 Astra leads
Claude Haiku 4.5: 49.9 (#129), GPT-6 Astra: 53.7 (#61)
| Benchmark | Claude Haiku 4.5 | GPT-6 Astra |
|---|---|---|
| LMArena Non-English | 1377 | 1430 |
| LMArena Chinese | 1417 | 1484 |
| LMArena French | 1408 | 1456 |
| LMArena German | 1375 | 1440 |
| LMArena Japanese | 1339 | 1379 |
| LMArena Korean | 1347 | 1426 |
| LMArena Russian | 1381 | 1436 |
| LMArena Spanish | 1420 | 1407 |
Instruction Following GPT-6 Astra leads
Claude Haiku 4.5: 71.4 (#149), GPT-6 Astra: 76.3 (#44)
| Benchmark | Claude Haiku 4.5 | GPT-6 Astra |
|---|---|---|
| LMArena Instruction Following | 1414 | 1450 |
| IFEval | 80.1% | — |
Long Context Too close to call
Claude Haiku 4.5: 43.6 (#92), GPT-6 Astra: 44.5 (#62)
| Benchmark | Claude Haiku 4.5 | GPT-6 Astra |
|---|---|---|
| LMArena Longer Query | 1427 | 1456 |
Writing & Preference GPT-6 Astra leads
Claude Haiku 4.5: 57.9 (#123), GPT-6 Astra: 75.3 (#7)
| Benchmark | Claude Haiku 4.5 | GPT-6 Astra |
|---|---|---|
| LMArena Text | 1396 | 1441 |
| LMArena Creative Writing | 1372 | 1418 |
| LMArena Multi-Turn | 1409 | 1448 |
| EQ-Bench Creative Writing | — | 2173 |
| WildBench | 83.9% | — |
| EQ-Bench 4 | 1064 | — |
Frequently asked questions
Is Claude Haiku 4.5 better than GPT-6 Astra?
GPT-6 Astra is the stronger model overall, scoring 70.8 to 39.5 on the Noometry Index. Claude Haiku 4.5 costs 10× less per token, which makes it the better buy when GPT-6 Astra's lead doesn't matter for your workload.
Which is cheaper, Claude Haiku 4.5 or GPT-6 Astra?
Claude Haiku 4.5 is cheaper. It lists at $1 per million input tokens and $5 per million output tokens; GPT-6 Astra lists at $10 and $50.
Is Claude Haiku 4.5 or GPT-6 Astra better for coding?
GPT-6 Astra scores higher on coding benchmarks: 73.7 versus 44.0 in the Noometry coding category.
Which has the bigger context window?
GPT-6 Astra does, with 1.05M tokens against 200K.
How many benchmarks do Claude Haiku 4.5 and GPT-6 Astra share?
37 benchmarks have published results for both models. Claude Haiku 4.5 has 53 scored results on Noometry and GPT-6 Astra has 56.