Compare / Gemma 4 31B vs gpt-oss-120b
Gemma 4 31B vs gpt-oss-120b
Which is cheaper, which is more capable, and which does better at each kind of work, from 26 results on 4 sources that measured both models.
Of the 26 results both models have, the better value on each. Results whose reported ranges overlap are too close to call.
Results as of 16 November 2026 (catalogue release 2026-11-16-49bdfa045dee). See also Google vs OpenAI, every model from both labs.
Gemma 4 31B or gpt-oss-120b? The short answer
Pick Gemma 4 31B: it scores higher (Intelligence Index 14.7 against 11.6) at about the same price. Pick gpt-oss-120b only if speed matters more: it writes faster.
- They cost about the same: $0.21 and $0.20 per million tokens (three input to one output).
- Gemma 4 31B scores higher on the Artificial Analysis Intelligence Index: 14.7 against 11.6.
- gpt-oss-120b writes faster: 259 tokens a second against 48.
- Gemma 4 31B is newer: released 2 Apr 2026, against 5 Aug 2025 for gpt-oss-120b.
- On third-party benchmarks, on the results both models have: Gemma 4 31B does better for agents and tool use, reasoning and knowledge, writing and sticking to the facts.
Head to head
Every fact side by side
| Gemma 4 31B | gpt-oss-120b | |
|---|---|---|
| Price per million tokens | $0.15 in · $0.40 out | $0.10 in · $0.50 out |
| Blended (3 in : 1 out) | $0.21 | $0.20 |
| Cheapest host, blended | $0.16 (CoreWeave, FP4) | $0.060 (Near AI, FP4) |
| Intelligence Index | 14.7 | 11.6 |
| Spring Prompt overall | – | – |
| Output speed, tokens a second | 48 | 259 |
| Context window | 262,144 tokens | 131,072 tokens |
| Released | 2 Apr 2026 (newer) | 5 Aug 2025 |
| Weights | Open | Open |
| Developer based in | the US | the US |
Prices are the developer's list price where we have read it, otherwise the typical host on OpenRouter; see every model's price. Intelligence and speed from Artificial Analysis.
Which is better for what
Each model's rank among the models on sale today, for each kind of work, from our own benchmarks and licensed sources. Further right is better; the better rank is in green.
Every shared result
Each model's best result on every measure both have, grouped by benchmark, with the setting that produced it. Under each measure, every model that benchmark has measured is a grey tick, best to the right, so you can see where the two sit in the field. The headline measures of each benchmark come first; the rest are under "Show all". Green marks the better value; where a source reports ranges that overlap, neither is marked.
| Measure | Gemma 4 31B | gpt-oss-120b |
|---|---|---|
| Arena (formerly LMArena) reported by Arena (formerly LMArena) · 3 results | ||
| Overall · TextArena rating, higher is better 65 · 187 of 387 measured | 1,453 | 1,352 |
| Overall · Text factualityArena rating, higher is better 60 · 142 of 163 measured | 1,450 | 1,385 |
| Business, management and finance · TextArena rating, higher is better 66 · 183 of 380 measured | 1,449 | 1,352 |
| Artificial Analysis reported by Artificial Analysis · 3 results | ||
| Artificial Analysis Intelligence Indexindex score, higher is better 136 · 162 of 249 measured | 14.7 | 11.6high reasoning |
| GPQA Diamond% of questions, higher is better 76 · 119 of 232 measured | 85.7% | 78.2%high reasoning |
| Humanity's Last Exam% of questions, higher is better 105 · 118 of 250 measured | 23.6% | 19.6%high reasoning |
| UGI Leaderboard reported by UGI Leaderboard · 3 results | ||
| Writing score · UGIscore out of 100, higher is better 94 · 116 of 230 measured | 43.1channel-thought-prefill reasoning | 38.5medium reasoning |
| Requested-length error · UGI% off the requested word count, lower is better 15 · 224 of 230 measured | 5.0% | 107.0%medium reasoning |
| Style adherence · UGIscore from 0 to 1, higher is better 73 · 217 of 230 measured | 0.36channel-thought-prefill reasoning | 0.29medium reasoning |
| Vectara Hallucination Leaderboard reported by Vectara Hallucination Leaderboard · 2 results | ||
| Hallucination rate · Vectara% of summaries, lower is better 35 · 88 of 104 measured | 7.4% | 14.2% |
| Answer rate · Vectara% of documents, higher is better 1 · 17 of 104 measured | 100.0% | 99.9% |
Show all 26 results (15 more)
| Measure | Gemma 4 31B | gpt-oss-120b |
|---|---|---|
| Arena (formerly LMArena) reported by Arena (formerly LMArena) · 4 results | ||
| Creative writing · TextArena rating, higher is better 74 · 238 of 385 measured | 1,420 | 1,278 |
| Expert prompts · TextArena rating, higher is better 42 · 188 of 338 measured | 1,498 | 1,357 |
| Instruction following · TextArena rating, higher is better 49 · 204 of 387 measured | 1,453 | 1,323 |
| Writing, literature and language · TextArena rating, higher is better 63 · 215 of 386 measured | 1,435 | 1,309 |
| Artificial Analysis reported by Artificial Analysis · 11 results | ||
| Terminal-Bench 2.1% of tasks, higher is better 72 · 88 of 131 measured | 43.4% | 26.2%high reasoning |
| Terminal-Bench 4.0% of tasks, higher is better 72 · 72 of 120 measured | 0.0% | 0.0%high reasoning |
| τ-bench banking% of tasks, higher is better 61 · 67 of 121 measured | 14.8% | 12.8%high reasoning |
| Artificial Analysis Coding Indexindex score, higher is better 69 · 85 of 130 measured | 43.4 | 30.4high reasoning |
| IFBench% of instructions, higher is better 24 · 58 of 195 measured | 75.6% | 69.0%high reasoning |
| Long-context reasoning (AA-LCR)% of questions, higher is better 119 · 160 of 241 measured | 69.7% | 52.0%high reasoning |
| Output speedtokens per second, higher is better 68 · 8 of 78 measured | 47.6no reasoning | 259high reasoning |
| SciCode% of problems, higher is better 72 · 101 of 121 measured | 45.5% | 34.0%high reasoning |
| Terminal-Bench Hard% of tasks, higher is better 46 · 84 of 190 measured | 36.4% | 23.5%high reasoning |
| Time to first answer tokenseconds, lower is better 22 · 45 of 78 measured | 0.77no reasoning | 8.22high reasoning |
| τ²-bench telecom% of tasks, higher is better 88 · 86 of 193 measured | 65.5%no reasoning | 65.8%high reasoning |
Compare them with other models
Quick answers
Is Gemma 4 31B better than gpt-oss-120b?
Pick Gemma 4 31B: it scores higher (Intelligence Index 14.7 against 11.6) at about the same price. Pick gpt-oss-120b only if speed matters more: it writes faster.
Which is cheaper, Gemma 4 31B or gpt-oss-120b?
They cost about the same: $0.21 and $0.20 per million tokens (three input to one output).
Which is faster, Gemma 4 31B or gpt-oss-120b?
gpt-oss-120b writes about 259 tokens a second on its usual API, against 48 for Gemma 4 31B (Artificial Analysis).
Which is better for agents and tool use, Gemma 4 31B or gpt-oss-120b?
Gemma 4 31B ranks 106 of 159 models on sale for multi-step tasks with tools: support desks, coding agents, function calls, against 125 for gpt-oss-120b.
Which is better for reasoning and knowledge, Gemma 4 31B or gpt-oss-120b?
Gemma 4 31B ranks 99 of 187 models on sale for hard questions across science, maths and general knowledge, against 118 for gpt-oss-120b.
Which is better for writing, Gemma 4 31B or gpt-oss-120b?
Gemma 4 31B ranks 72 of 169 models on sale for what people prefer in blind comparisons, and judged writing quality, against 124 for gpt-oss-120b.
Which is better for sticking to the facts, Gemma 4 31B or gpt-oss-120b?
Gemma 4 31B ranks 85 of 139 models on sale for summarising without inventing things, and factual answers, against 129 for gpt-oss-120b.
Which has the bigger context window, Gemma 4 31B or gpt-oss-120b?
Gemma 4 31B: 262,144 tokens, against 131,072 for gpt-oss-120b.
Run Gemma 4 31B and gpt-oss-120b on your own prompt.
Benchmarks aren't your task. Test both side by side on your prompt, with every answer checked against your rules and the cost of each one.