Model comparison
Claude Haiku 4.5 vs gpt-oss-20b
Claude Haiku 4.5 is the stronger model overall, scoring 39.5 to 32.5 on the Noometry Index. gpt-oss-20b costs 56× less per token, which makes it the better buy when Claude Haiku 4.5's lead doesn't matter for your workload.
Last verified . 32 shared benchmarks.
Summary
- They share 32 benchmarks with published results for both. Claude Haiku 4.5 scores higher in 8 categories and gpt-oss-20b in 1 category; 9 gaps are clear of the uncertainty.
- The widest gap is in agentic & tool use, where Claude Haiku 4.5 leads 33.6 to 9.3.
- The biggest single-benchmark swing is Terminal-Bench: 35.5% for Claude Haiku 4.5 and 3.4% for gpt-oss-20b.
- gpt-oss-20b is cheaper at $0.018 / $0.09 per million input/output tokens, against $1 / $5 for Claude Haiku 4.5.
- Claude Haiku 4.5 accepts more context: 200K tokens versus 131K.
- gpt-oss-20b has downloadable open weights; the other is API-only.
Side by side
| Claude Haiku 4.5 | gpt-oss-20b | |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 39.5 | 32.5 |
| Released | 2025-10-15 | 2025-08-05 |
| Weights | Proprietary | Open |
| Context window | 200K | 131K |
| Max output | 64K | 16K |
| Input $ / M tokens | $1 | $0.018 |
| Output $ / M tokens | $5 | $0.09 |
| Results tracked | 53 | 34 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Claude Haiku 4.5 leads
Claude Haiku 4.5: 44.0 (#78), gpt-oss-20b: 37.6 (#192)
| Benchmark | Claude Haiku 4.5 | gpt-oss-20b |
|---|---|---|
| SciCode | 43.3% | 34.4% |
| WeirdML | 45.4% | 40.9% |
| LMArena Coding | 1453 | 1306 |
| ALE-Bench | 653.48 | 566.05 |
| SWE-bench Verified (bash only) | 66.6% | — |
| LMArena WebDev | 1330 | — |
| SWE-bench Multilingual | 64.7% | — |
Agentic & Tool Use Claude Haiku 4.5 leads
Claude Haiku 4.5: 33.6 (#52), gpt-oss-20b: 9.3 (#154)
| Benchmark | Claude Haiku 4.5 | gpt-oss-20b |
|---|---|---|
| Terminal-Bench | 35.5% | 3.4% |
| Berkeley Function Calling Leaderboard | 68.7% | — |
| DeepResearch Bench | 45.5% | — |
| BALROG | 31.2% | — |
| ExploitBench | 13.7% | — |
| Vending-Bench 2 | 458.89 | — |
Reasoning gpt-oss-20b leads
Claude Haiku 4.5: 15.1 (#320), gpt-oss-20b: 19.3 (#261)
| Benchmark | Claude Haiku 4.5 | gpt-oss-20b |
|---|---|---|
| CritPt | 0% | 1.4% |
| Chess Puzzles | 8% | 4% |
| LMArena Hard Prompts | 1420 | 1274 |
| DTBench | 73.6% | 68% |
| LMCA | 30.9% | 14.5% |
| Epoch Capabilities Index | 142.41 | 137.82 |
| ARC-AGI-2 | 4% | — |
| Kagi LLM Benchmark | — | 53.2% |
| NYT Connections (extended) | 14.3% | — |
| ARC-AGI-1 | 47.7% | — |
| ForecastBench | 61.4 | — |
Math Claude Haiku 4.5 leads
Claude Haiku 4.5: 44.9 (#78), gpt-oss-20b: 39.4 (#103)
| Benchmark | Claude Haiku 4.5 | gpt-oss-20b |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 66.7% | 65.3% |
| Omni-MATH | 56.1% | 56.5% |
| LMArena Math | 1396 | 1317 |
| MATH Level 5 | 96.4% | — |
| FrontierMath (Feb 2025 set) | 5.9% | — |
| FrontierMath Tier 4 (v1) | 2.1% | — |
Knowledge Claude Haiku 4.5 leads
Claude Haiku 4.5: 37.7 (#153), gpt-oss-20b: 34.6 (#195)
| Benchmark | Claude Haiku 4.5 | gpt-oss-20b |
|---|---|---|
| GPQA Diamond | 71.2% | 60.8% |
| MMLU-Pro | 77.7% | 74% |
| GPQA (HELM) | 60.5% | 59.4% |
| LMArena Expert | 1442 | 1258 |
| SimpleQA Verified | 13.2% | — |
| Vectara Hallucination Rate | 9.8% | — |
Multimodal Not comparable
Claude Haiku 4.5: 26.8 (#118), gpt-oss-20b: —
| Benchmark | Claude Haiku 4.5 | gpt-oss-20b |
|---|---|---|
| Blueprint-Bench 2 | 0% | — |
| LMArena Document | 1420 | — |
Multilingual Claude Haiku 4.5 leads
Claude Haiku 4.5: 49.9 (#129), gpt-oss-20b: 42.2 (#197)
| Benchmark | Claude Haiku 4.5 | gpt-oss-20b |
|---|---|---|
| LMArena Non-English | 1377 | 1268 |
| LMArena Chinese | 1417 | 1314 |
| LMArena German | 1375 | 1255 |
| LMArena Japanese | 1339 | 1244 |
| LMArena Korean | 1347 | 1236 |
| LMArena Russian | 1381 | 1278 |
| LMArena Spanish | 1420 | 1267 |
| LMArena French | 1408 | — |
Instruction Following Claude Haiku 4.5 leads
Claude Haiku 4.5: 71.4 (#149), gpt-oss-20b: 61.8 (#240)
| Benchmark | Claude Haiku 4.5 | gpt-oss-20b |
|---|---|---|
| IFEval | 80.1% | 73.2% |
| LMArena Instruction Following | 1414 | 1236 |
Long Context Claude Haiku 4.5 leads
Claude Haiku 4.5: 43.6 (#92), gpt-oss-20b: 37.9 (#209)
| Benchmark | Claude Haiku 4.5 | gpt-oss-20b |
|---|---|---|
| LMArena Longer Query | 1427 | 1250 |
Writing & Preference Claude Haiku 4.5 leads
Claude Haiku 4.5: 57.9 (#123), gpt-oss-20b: 35.5 (#265)
| Benchmark | Claude Haiku 4.5 | gpt-oss-20b |
|---|---|---|
| LMArena Text | 1396 | 1287 |
| LMArena Creative Writing | 1372 | 1201 |
| WildBench | 83.9% | 73.7% |
| LMArena Multi-Turn | 1409 | 1268 |
| EQ-Bench Creative Writing | — | 666 |
| EQ-Bench 4 | 1064 | — |
Frequently asked questions
Is Claude Haiku 4.5 better than gpt-oss-20b?
Claude Haiku 4.5 is the stronger model overall, scoring 39.5 to 32.5 on the Noometry Index. gpt-oss-20b costs 56× less per token, which makes it the better buy when Claude Haiku 4.5's lead doesn't matter for your workload.
Which is cheaper, Claude Haiku 4.5 or gpt-oss-20b?
gpt-oss-20b is cheaper. It lists at $0.018 per million input tokens and $0.09 per million output tokens; Claude Haiku 4.5 lists at $1 and $5.
Is Claude Haiku 4.5 or gpt-oss-20b better for coding?
Claude Haiku 4.5 scores higher on coding benchmarks: 44.0 versus 37.6 in the Noometry coding category.
Which has the bigger context window?
Claude Haiku 4.5 does, with 200K tokens against 131K.
How many benchmarks do Claude Haiku 4.5 and gpt-oss-20b share?
32 benchmarks have published results for both models. Claude Haiku 4.5 has 53 scored results on Noometry and gpt-oss-20b has 34.