DeepSeek V3 0324 vs GPT-4.1 mini vs Kimi K2 Instruct
GPT-4.1 mini comes out ahead, 64 to 55 and 52 on our weighted score, though DeepSeek V3 0324 is 31% cheaper per token.
Price in rating: Excluded · Include price
DeepSeek
DeepSeek V3 0324
52/100- ECI135.3
- Price$0.275 / $1.12
- Context164K
- Our pick
OpenAI
GPT-4.1 mini
64/100- ECI134.3
- Price$0.40 / $1.60
- Context1.05M
Moonshot AI
Kimi K2 Instruct
55/100- ECI141.3
- Price$0.57 / $2.30
- Context131K
GPT-4.1 mini is our pick
GPT-4.1 mini is the better all-round choice, scoring 64/100 against Kimi K2 Instruct (55) and DeepSeek V3 0324 (52). It leads on inputs & features and context window. Kimi K2 Instruct wins on capability. The score weighs capability 67%, inputs & features 20%, context window 13%.
- CapabilityKimi K2 InstructCapabilities Index (ECI): Kimi K2 Instruct 141.3 · DeepSeek V3 0324 135.3 · GPT-4.1 mini 134.3
- Lowest priceDeepSeek V3 0324DeepSeek V3 0324 $0.486 · GPT-4.1 mini $0.70 · Kimi K2 Instruct $1.00 per 1M tokens (3:1 blend)
- Longest contextGPT-4.1 miniGPT-4.1 mini 1,047,576 · DeepSeek V3 0324 163,840 · Kimi K2 Instruct 131,072 tokens
- Widest inputsGPT-4.1 miniDeepSeek V3 0324: Text · GPT-4.1 mini: Text, Images, PDFs · Kimi K2 Instruct: Text
- Self-hostingDeepSeek V3 0324 and Kimi K2 InstructPublishes downloadable weights (Modified MIT)
| Measure | Weight | DeepSeek V3 0324 | GPT-4.1 mini | Kimi K2 Instruct |
|---|---|---|---|---|
| CapabilityCapabilities Index (ECI) | 67% | 63 | 62 | 69 |
| Inputs & features | 20% | 25 | 70 | 25 |
| Context window | 13% | 36 | 65 | 32 |
| Overall | 100% | 52/100 | 64/100 | 55/100 |
Every spec in one table
Highlighted cells lead their row. Dashes mean the data is not published.
| Specification | |||
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | 135.3 | 134.3 | 141.3 (best) |
| ECI rank | #125 of 158 | #128 of 158 | #108 of 158 (best) |
| GPQA DiamondGraduate-level science questions | 67.6% (best) | 65.9% | — |
| FrontierMath Tiers 1–3Research-level mathematics | — | 6.7% | — |
| OTIS Mock AIME 2024–2025Competition mathematics | 37.8% | 44.7% (best) | — |
| SimpleQA VerifiedShort factual questions | — | 12.7% | — |
| Price per million tokens | |||
| Input | $0.275 (best) | $0.40 | $0.57 |
| Output | $1.12 (best) | $1.60 | $2.30 |
| Cached input | — | $0.10 | — |
| Blended (3:1) | $0.486 (best) | $0.70 | $1.00 |
| Long-context rate | Same rate | Same rate | Same rate |
| Price source | Median of 13 providers | Official OpenAI API | Median of 11 providers |
| Limits | |||
| Context window | 163,840 tokens | 1,047,576 tokens (best) | 131,072 tokens |
| Max output | 163,840 tokens (best) | 32,768 tokens | 32,768 tokens |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | No | Yes | No |
| PDFs | No | Yes | No |
| Audio | No | No | No |
| Video | No | No | No |
| Reasoning | No | No | No |
| Tool calling | Yes | Yes | Yes |
| Structured output | No | Yes | No |
| Availability | |||
| Weights | Open | Proprietary | OpenModified MIT |
| API model ID | — | gpt-4.1-mini | — |
| API providers | 13 | 30 (best) | 11 |
| Released | Mar 24, 2025 | Apr 14, 2025 | Jul 11, 2025 |
| Knowledge cutoff | — | 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.
DeepSeek V3 0324$4.99
GPT-4.1 mini$7.20
Kimi K2 Instruct$10.30
Which should you choose?
Which is better: DeepSeek V3 0324, GPT-4.1 mini or Kimi K2 Instruct?
GPT-4.1 mini is the better all-round choice, scoring 64/100 against Kimi K2 Instruct (55) and DeepSeek V3 0324 (52). It leads on inputs & features and context window. Kimi K2 Instruct wins on capability. The score weighs capability 67%, inputs & features 20%, context window 13%.
Which is cheaper, DeepSeek V3 0324, GPT-4.1 mini or Kimi K2 Instruct?
DeepSeek V3 0324 is cheaper at $0.275 input / $1.12 output per million tokens (median across 13 API providers). GPT-4.1 mini costs $0.40 input / $1.60 output per million tokens (official OpenAI API price); Kimi K2 Instruct costs $0.57 input / $2.30 output per million tokens (median across 11 API providers). At a typical mix of three input tokens to one output token, that is $0.486 per million tokens for DeepSeek V3 0324 versus $0.70 for GPT-4.1 mini (1.4× as much) and $1.00 for Kimi K2 Instruct (2.1× as much).
Which scores higher on benchmarks?
Kimi K2 Instruct scores higher on the Capabilities Index (ECI): Kimi K2 Instruct 141.3 (#108 of 158), DeepSeek V3 0324 135.3 (#125 of 158) and GPT-4.1 mini 134.3 (#128 of 158). Their confidence ranges do not overlap (138.3–144.0 vs 133.1–137.3), so the gap is a real one.
Which is better for coding?
There are no published SWE-bench Verified results for DeepSeek V3 0324, GPT-4.1 mini and Kimi K2 Instruct yet, so there is no like-for-like coding score. On overall capability, Kimi K2 Instruct 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?
GPT-4.1 mini has the largest context window at 1,047,576 tokens, against 163,840 for DeepSeek V3 0324 and 131,072 for Kimi K2 Instruct. Maximum output per response: DeepSeek V3 0324 up to 163,840, GPT-4.1 mini up to 32,768, Kimi K2 Instruct up to 32,768 tokens.
Which can read images, PDFs, audio or video?
DeepSeek V3 0324 accepts text; GPT-4.1 mini accepts text, images and PDFs; Kimi K2 Instruct accepts text. GPT-4.1 mini handles the widest range of inputs.
Are any of these open source?
DeepSeek V3 0324 and Kimi K2 Instruct publishes its weights (Modified MIT) and can be self-hosted; GPT-4.1 mini is proprietary.
Which is newer?
Kimi K2 Instruct is the newest, released Jul 11, 2025. GPT-4.1 mini came out Apr 14, 2025; DeepSeek V3 0324 came out Mar 24, 2025. Knowledge cutoff: GPT-4.1 mini 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.