Gemma 3 4B IT vs Mistral Nemo vs Qwen2.5 7B Instruct
Too close to call on our weighted score (Gemma 3 4B IT 52, Qwen2.5 7B Instruct 50, Mistral Nemo 47). The right pick depends on what you value most.
GoogleGemma 3 4B IT
52/100- ECI115.8
- Price$0.05 / $0.10
- Context131K
Mistral AIMistral Nemo
47/100- ECI118.6
- Price$0.12 / $0.13
- Context128K
Alibaba (Qwen)Qwen2.5 7B Instruct
50/100- ECI118.5
- Price$0.05 / $0.10
- Context131K
Too close to call
It is close. Our weighted score puts them within 2 points (Gemma 3 4B IT 52/100, Qwen2.5 7B Instruct 50/100, Mistral Nemo 47/100), so choose by what matters most for your work: Mistral Nemo for raw capability and Gemma 3 4B IT on price. The score weighs capability 50%, price 25%, inputs & features 15%, context window 10%.
- CapabilityMistral NemoCapabilities Index (ECI): Mistral Nemo 118.6 · Qwen2.5 7B Instruct 118.5 · Gemma 3 4B IT 115.8
- Lowest priceGemma 3 4B IT and Qwen2.5 7B InstructGemma 3 4B IT $0.063 · Qwen2.5 7B Instruct $0.063 · Mistral Nemo $0.123 per 1M tokens (3:1 blend)
- Longest contextGemma 3 4B IT and Qwen2.5 7B InstructGemma 3 4B IT 131,072 · Qwen2.5 7B Instruct 131,072 · Mistral Nemo 128,000 tokens
- Widest inputsGemma 3 4B ITGemma 3 4B IT: Text, Images · Mistral Nemo: Text · Qwen2.5 7B Instruct: Text
- Self-hostingAll open weightsEvery model here can be downloaded and run on your own hardware
| Measure | Weight | Gemma 3 4B IT | Mistral Nemo | Qwen2.5 7B Instruct |
|---|---|---|---|---|
| CapabilityCapabilities Index (ECI) | 50% | 34 | 38 | 38 |
| Price | 25% | 100 | 89 | 100 |
| Inputs & features | 15% | 50 | 25 | 25 |
| Context window | 10% | 25 | 24 | 25 |
| Overall | 100% | 52/100 | 47/100 | 50/100 |
Every spec in one table
Highlighted cells lead their row. Dashes mean the data is not published.
| Specification | Gemma 3 4B ITGoogle | Mistral NemoMistral AI | Qwen2.5 7B InstructAlibaba (Qwen) |
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | 115.8 | 118.6 (best) | 118.5 |
| ECI rank | #163 of 165 | #157 of 165 (best) | #159 of 165 |
| GPQA DiamondGraduate-level science questions | 23.2% | 29.9% | 35.5% (best) |
| OTIS Mock AIME 2024–2025Competition mathematics | 7.5% (best) | — | 2.5% |
| Price per million tokens | |||
| Input | $0.05 (best) | $0.12 | $0.05 (best) |
| Output | $0.10 (best) | $0.13 | $0.10 (best) |
| Cached input | — | — | — |
| Blended (3:1) | $0.063 (best) | $0.123 | $0.063 (best) |
| Long-context rate | Same rate | Same rate | Same rate |
| Price source | Median of 9 providers | Median of 12 providers | Median of 3 providers |
| Limits | |||
| Context window | 131,072 tokens (best) | 128,000 tokens | 131,072 tokens (best) |
| Max output | 131,072 tokens (best) | 128,000 tokens | 8,192 tokens |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | Yes | No | No |
| PDFs | No | No | No |
| Audio | No | No | No |
| Video | No | No | No |
| Reasoning | No | No | No |
| Tool calling | Yes | Yes | Yes |
| Structured output | No | No | No |
| Availability | |||
| Weights | Open | Open | OpenApache 2.0 |
| API model ID | — | — | — |
| API providers | 10 | 12 (best) | 4 |
| Released | Mar 12, 2025 | Jul 18, 2024 | Sep 19, 2024 |
| Knowledge cutoff | Aug 2024 | Jul 2024 | — |
What would a month cost?
Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.
- Gemma 3 4B IT$0.70
- Mistral Nemo$1.46
- Qwen2.5 7B Instruct$0.70
Which should you choose?
Which is better: Gemma 3 4B IT, Mistral Nemo or Qwen2.5 7B Instruct?
It is close. Our weighted score puts them within 2 points (Gemma 3 4B IT 52/100, Qwen2.5 7B Instruct 50/100, Mistral Nemo 47/100), so choose by what matters most for your work: Mistral Nemo for raw capability and Gemma 3 4B IT on price. The score weighs capability 50%, price 25%, inputs & features 15%, context window 10%.
Which is cheaper, Gemma 3 4B IT, Mistral Nemo or Qwen2.5 7B Instruct?
Gemma 3 4B IT is cheaper at $0.05 input / $0.10 output per million tokens (median across 9 API providers). Qwen2.5 7B Instruct costs $0.05 input / $0.10 output per million tokens (median across 3 API providers); Mistral Nemo costs $0.12 input / $0.13 output per million tokens (median across 12 API providers). At a typical mix of three input tokens to one output token, that is $0.063 per million tokens for Gemma 3 4B IT versus $0.063 for Qwen2.5 7B Instruct (1× as much) and $0.123 for Mistral Nemo (2× as much).
Which scores higher on benchmarks?
Mistral Nemo scores higher on the Capabilities Index (ECI): Mistral Nemo 118.6 (#157 of 165), Qwen2.5 7B Instruct 118.5 (#159 of 165) and Gemma 3 4B IT 115.8 (#163 of 165). The confidence ranges of the top two overlap (111.6–121.6 vs 110.6–121.2), so treat the gap as small. On individual benchmarks: GPQA Diamond — Qwen2.5 7B Instruct 35.5%, Mistral Nemo 29.9%, Gemma 3 4B IT 23.2%.
Which is better for coding?
There are no published SWE-bench Verified results for Gemma 3 4B IT, Mistral Nemo and Qwen2.5 7B Instruct yet, so there is no like-for-like coding score. On overall capability, Mistral Nemo 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?
Gemma 3 4B IT and Qwen2.5 7B Instruct have the largest context windows (131,072 and 131,072 tokens), against 128,000 for Mistral Nemo. Maximum output per response: Gemma 3 4B IT up to 131,072, Mistral Nemo up to 128,000, Qwen2.5 7B Instruct up to 8,192 tokens.
Which can read images, PDFs, audio or video?
Gemma 3 4B IT accepts text and images; Mistral Nemo accepts text; Qwen2.5 7B Instruct accepts text. Gemma 3 4B IT handles the widest range of inputs.
Are any of these open source?
Yes, all three publish their weights (Apache 2.0), so you can self-host them.
Which is newer?
Gemma 3 4B IT is the newest, released Mar 12, 2025. Qwen2.5 7B Instruct came out Sep 19, 2024; Mistral Nemo came out Jul 18, 2024. Knowledge cutoff: Gemma 3 4B IT Aug 2024, Mistral Nemo Jul 2024.
How do you decide the winner?
Each model gets a 0–100 score on capability (50%, independent benchmark results); price (25%, blended price per million tokens (3 input : 1 output), log scale); inputs & features (15%, image, PDF, audio and video input, tool calling, structured output and reasoning); context window (10%, 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 2, 2026.
