Across 12 comparable benchmarks, gpt-oss-20b has better reported results on 10 and worse on 2.
Moderate evidence12 of 12 shared results countedHow we compare
Price basis: Standard rates vs listed rates. The price section below shows the applicable periods.
| Fact | gpt-oss-20b | Qwen3-30B-A3B-2507 |
|---|---|---|
| Context window | 128K tokens | 256K tokens |
| Input / 1M | $0.030 | — |
| Output / 1M | $0.140 | — |
| Blended (4:1) | $0.052 | — |
| Status | Available | Deprecated |
| Released | August 6, 2025 | July 29, 2025 |
Blended is 4:1 input to output with 90% cache hits at listed rates — set your own mix.
Mean of 7 shared benchmarks grouped by capability on a 0–100 scale. How averages are built
gpt-oss-20b ahead on 2 of 2 counted
gpt-oss-20b ahead on 1 of 1 counted
gpt-oss-20b ahead on 3 of 3 counted
gpt-oss-20b ahead on 1 of 1 counted
gpt-oss-20b ahead on 1 of 1 counted
Split 1–1 of 2 counted
No shared comparable results for Abstract reasoning, Computer & tool use, Long context, Cybersecurity.
Published scores, with matching modes where available. Gaps show gpt-oss-20b relative to Qwen3-30B-A3B-2507. Reading the results
| Benchmark | gpt-oss-20b | Qwen3-30B-A3B-2507 | Gap |
|---|---|---|---|
| SWE-bench VerifiedRepository Engineering · Accuracy | 34.0Thinking · no tools | 22.0Thinking · no tools | +12.0 pts#103 vs #107 of 108 |
| Terminal Bench HardAgentic Development · Accuracy | 10.6Thinking High · tools | 5.3≠Thinking · tools | +5.3 pts#115 vs #135 of 157 |
| AIME2025Mathematics · Accuracy | 98.7Thinking · tools | 61.3≠Normal · no tools | +37.4 pts#13 vs #75 of 86 |
| CritPtScientific Reasoning · Score | 1.4Thinking High · no tools | 0.3≠Thinking · no tools | 4.67×#96 vs #124 of 134 |
| GPQA DiamondScientific Reasoning · Accuracy | 71.5Thinking · no tools | 70.4≠Normal · no tools | +1.1 pts#130 vs #138 of 200 |
| SciCodeScientific Computing · Score | 38.9Thinking High · no tools | 33.0≠Thinking · no tools | 1.18×#79 vs #89 of 98 |
| BrowseCompFact Finding · Accuracy | 28.3Thinking · tools | 2.3Thinking · tools | +26.0 pts#58 vs #61 of 61 |
| GDP.pdfDocuments & Charts · Task pass rate (%) | 2.0Thinking High · no tools | 0.4≠Thinking · no tools | +1.6 pts#84 vs #96 of 96 |
| τ²-BenchService Workflows · Accuracy | 47.7Thinking · tools | 49.0Thinking · tools | −1.3 pts#34 vs #33 of 40 |
| τ³-BankingService Workflows · Score | 7.0Thinking High · tools | 5.4≠Thinking · tools | 1.30×#99 vs #107 of 120 |
| HLEKnowledge Exams · Accuracy | 17.3Thinking · tools | 9.8≠Thinking · no tools | +7.5 pts#114 vs #128 of 151 |
| MMLU-ProKnowledge Exams · Accuracy | 74.0Thinking · no tools | 78.4≠Normal · no tools | −4.4 pts#84 vs #69 of 153 |
Repository Engineering · Accuracy
Agentic Development · Accuracy
Mathematics · Accuracy
Scientific Reasoning · Score
Scientific Reasoning · Accuracy
Scientific Computing · Score
Fact Finding · Accuracy
Documents & Charts · Task pass rate (%)
Service Workflows · Accuracy
Service Workflows · Score
A dash means no result is available.
| Benchmark | gpt-oss-20b | Qwen3-30B-A3B-2507 |
|---|---|---|
| AIME 2024 | 96.0 | — |
| MMLU | 85.3 | — |
| CodeForces | 2516 | — |
| LiveCodeBench | — | 43.2 |
| Pinch Bench | 66.0 | — |
| PinchBench v2 | 36.3 | — |
| ECI | 138 | — |
| Creative Writing | 665 | — |
Time-of-day providers use peak-hour rates; fixed-price providers use standard rates. Off-peak discounts remain in the published details.
Drag the handle above the bar for cache hits, the one below it for the input : output mix · 5% steps
Blended price = effective input rate at 90% cache hits × 80% + output rate × 20%
A blended cost needs a published input and output price on both sides, and one of these two does not have them yet.
gpt-oss-20b: Standard pricing. DeepInfra official models API (openai/gpt-oss-20b) verified 2026-11-13
Qwen3-30B-A3B-2507: Listed text rates; pricing tier not confirmed.
| Price / 1M tokens | gpt-oss-20b | Qwen3-30B-A3B-2507 | Ratio |
|---|---|---|---|
| Input | $0.03 | — | — |
| Output | $0.14 | — | — |
| Cached input | — | — | — |
| Cache write | — | — | — |
| Blended | $0.052 | — | — |
Estimates use the selected token mix. Cache writes, tool fees and long-context surcharges are excluded.
| Attribute | gpt-oss-20b | Qwen3-30B-A3B-2507 |
|---|---|---|
| Availability | Available | Deprecated |
| Provider | OpenAI | Alibaba |
| Model type | Reasoning model | Chat model |
| First release | August 6, 2025 | July 29, 2025 |
| Context window | 128K tokens | 256K tokens |
| Max output | 128K tokens | 16K tokens |
| Total parameters | 21B | 30.5B |
| Active parameters | 3.6B | 3.3B |
| Architecture | Mixture of Experts (MoE) | Mixture of Experts (MoE) |
| Documented capabilities | Reasoning, Multilingual | Multilingual |
| Input modalities | Text | Text |
| Output modalities | Text | Text |
| Weights | Open weights | Open weights |
| Licence | Apache 2.0 | Apache 2.0 |
| Commercial use | Commercial use allowed | Commercial use allowed |
These are reported-score comparisons, not controlled head-to-head tests.
What people found comparing these models, or using either one. Comments also appear on the models' own pages.
Used gpt-oss-20b or Qwen3-30B-A3B-2507? Sign in with Google to share what worked, what did not, and which API or setup you tested.
No community comments on either model yet.
Alternatives with the most benchmark overlap for this matchup.