GPT OSS 20B vs Qwen3 30B-A3B Instruct 2507 vs QwQ 32B
Too close to call on our weighted score (QwQ 32B 54, GPT OSS 20B 53, Qwen3 30B-A3B Instruct 2507 53). The right pick depends on what you value most.
Price in rating: Excluded · Include price
OpenAI
GPT OSS 20B
53/100- ECI134.9
- Price$0.05 / $0.20
- Context131K
Alibaba (Qwen)
Qwen3 30B-A3B Instruct 2507
53/100- ECI134.8
- Price$0.10 / $0.30
- Context262K
Alibaba (Qwen)
QwQ 32B
54/100- ECI136.5
- Price$0.53 / $0.70
- Context131K
Too close to call
It is close. Our weighted score puts them within 1 points (QwQ 32B 54/100, GPT OSS 20B 53/100, Qwen3 30B-A3B Instruct 2507 53/100), so choose by what matters most for your work: QwQ 32B for raw capability, GPT OSS 20B on price and Qwen3 30B-A3B Instruct 2507 for long inputs. The score weighs capability 67%, inputs & features 20%, context window 13%.
- CapabilityQwQ 32BCapabilities Index (ECI): QwQ 32B 136.5 · GPT OSS 20B 134.9 · Qwen3 30B-A3B Instruct 2507 134.8
- Lowest priceGPT OSS 20BGPT OSS 20B $0.087 · Qwen3 30B-A3B Instruct 2507 $0.15 · QwQ 32B $0.573 per 1M tokens (3:1 blend)
- Longest contextQwen3 30B-A3B Instruct 2507Qwen3 30B-A3B Instruct 2507 262,144 · GPT OSS 20B 131,072 · QwQ 32B 131,072 tokens
- Widest inputsSame inputsGPT OSS 20B: Text · Qwen3 30B-A3B Instruct 2507: Text · QwQ 32B: Text
- Self-hostingAll open weightsEvery model here can be downloaded and run on your own hardware
| Measure | Weight | GPT OSS 20B | Qwen3 30B-A3B Instruct 2507 | QwQ 32B |
|---|---|---|---|---|
| CapabilityCapabilities Index (ECI) | 67% | 63 | 63 | 65 |
| Inputs & features | 20% | 35 | 25 | 35 |
| Context window | 13% | 32 | 43 | 32 |
| Overall | 100% | 53/100 | 53/100 | 54/100 |
Every spec in one table
Highlighted cells lead their row. Dashes mean the data is not published.
| Specification | |||
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | 134.9 | 134.8 | 136.5 (best) |
| ECI rank | #126 of 158 | #127 of 158 | #122 of 158 (best) |
| GPQA DiamondGraduate-level science questions | 60.8% | 55.6% | 65.3% (best) |
| OTIS Mock AIME 2024–2025Competition mathematics | 65.3% (best) | 62.2% | 59.2% |
| Price per million tokens | |||
| Input | $0.05 (best) | $0.10 | $0.53 |
| Output | $0.20 (best) | $0.30 | $0.70 |
| Cached input | — | — | — |
| Blended (3:1) | $0.087 (best) | $0.15 | $0.573 |
| Long-context rate | Same rate | Same rate | Same rate |
| Price source | Median of 39 providers | Median of 12 providers | Median of 2 providers |
| Limits | |||
| Context window | 131,072 tokens | 262,144 tokens (best) | 131,072 tokens |
| Max output | 131,072 tokens (best) | 16,384 tokens | 8,192 tokens |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | No | No | No |
| PDFs | No | No | No |
| Audio | No | No | No |
| Video | No | No | No |
| Reasoning | Yes | No | Yes |
| Tool calling | Yes | Yes | Yes |
| Structured output | No | No | No |
| Availability | |||
| Weights | Open | OpenApache 2.0 | Open |
| API model ID | — | — | — |
| API providers | 44 (best) | 14 | 2 |
| Released | Aug 5, 2025 | Jul 28, 2025 | Mar 5, 2025 |
| Knowledge cutoff | Jun 2024 | — | Apr 2024 |
What would a month cost?
Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.
GPT OSS 20B$0.90
Qwen3 30B-A3B Instruct 2507$1.60
QwQ 32B$6.70
Which should you choose?
Which is better: GPT OSS 20B, Qwen3 30B-A3B Instruct 2507 or QwQ 32B?
It is close. Our weighted score puts them within 1 points (QwQ 32B 54/100, GPT OSS 20B 53/100, Qwen3 30B-A3B Instruct 2507 53/100), so choose by what matters most for your work: QwQ 32B for raw capability, GPT OSS 20B on price and Qwen3 30B-A3B Instruct 2507 for long inputs. The score weighs capability 67%, inputs & features 20%, context window 13%.
Which is cheaper, GPT OSS 20B, Qwen3 30B-A3B Instruct 2507 or QwQ 32B?
GPT OSS 20B is cheaper at $0.05 input / $0.20 output per million tokens (median across 39 API providers). Qwen3 30B-A3B Instruct 2507 costs $0.10 input / $0.30 output per million tokens (median across 12 API providers); QwQ 32B costs $0.53 input / $0.70 output per million tokens (median across 2 API providers). At a typical mix of three input tokens to one output token, that is $0.087 per million tokens for GPT OSS 20B versus $0.15 for Qwen3 30B-A3B Instruct 2507 (1.7× as much) and $0.573 for QwQ 32B (6.5× as much).
Which scores higher on benchmarks?
QwQ 32B scores higher on the Capabilities Index (ECI): QwQ 32B 136.5 (#122 of 158), GPT OSS 20B 134.9 (#126 of 158) and Qwen3 30B-A3B Instruct 2507 134.8 (#127 of 158). The confidence ranges of the top two overlap (133.7–139.2 vs 132.3–137.3), so treat the gap as small. On individual benchmarks: GPQA Diamond — QwQ 32B 65.3%, GPT OSS 20B 60.8%, Qwen3 30B-A3B Instruct 2507 55.6%; OTIS Mock AIME 2024–2025 — GPT OSS 20B 65.3%, Qwen3 30B-A3B Instruct 2507 62.2%, QwQ 32B 59.2%.
Which is better for coding?
There are no published SWE-bench Verified results for GPT OSS 20B, Qwen3 30B-A3B Instruct 2507 and QwQ 32B yet, so there is no like-for-like coding score. On overall capability, QwQ 32B leads, which tends to carry over to coding, but test on your own codebase. All three support tool calling for agent workflows.
Which has the bigger context window?
Qwen3 30B-A3B Instruct 2507 has the largest context window at 262,144 tokens, against 131,072 for GPT OSS 20B and 131,072 for QwQ 32B. Maximum output per response: GPT OSS 20B up to 131,072, Qwen3 30B-A3B Instruct 2507 up to 16,384, QwQ 32B up to 8,192 tokens.
Which can read images, PDFs, audio or video?
GPT OSS 20B accepts text; Qwen3 30B-A3B Instruct 2507 accepts text; QwQ 32B accepts text. They handle the same number of input types.
Are any of these open source?
Yes, all three publish their weights (Apache 2.0), so you can self-host them.
Which is newer?
GPT OSS 20B is the newest, released Aug 5, 2025. Qwen3 30B-A3B Instruct 2507 came out Jul 28, 2025; QwQ 32B came out Mar 5, 2025. Knowledge cutoff: GPT OSS 20B Jun 2024, QwQ 32B Apr 2024.
How do you decide the winner?
This rating uses capability (67%, independent benchmark results); inputs & features (20%, image, PDF, audio and video input, tool calling, structured output and reasoning); context window (13%, maximum tokens per request, log scale). Dimensions missing for any model are dropped and the remaining weights rescaled, so every model is judged on the same evidence. Specs and prices come from public model listings and the labs’ own API pages; capability scores come from independent benchmark runs. Data updated Nov 18, 2026.