Kimi K3 vs DeepSeek V4 Pro
Side-by-side comparison of API pricing, specs, benchmarks, and capabilities
DeepSeek V4 Pro is 67% cheaper on blended cost ($1.98 vs $6.00/Mtok) — note: different model classes (mid tier vs flagship)
| Specification |
by Moonshot
|
by DeepSeek
4 providers:
DeepSeek $1.32/$3.96
Fireworks $1.74/$3.48
Together $1.74/$3.48
DeepInfra $1.30/$2.60
|
|---|---|---|
| Overview | ||
| Status | Current flagship Open weights | Current mid tier |
| Released | Jul 16, 2026 | Apr 24, 2026 |
| Pricing per million tokens | ||
| Input | $3.00/Mtok | $1.32/Mtok |
| Output | $15.00/Mtok | $3.96/Mtok |
| Blended avg | $6.00/Mtok | $1.98/Mtok |
| Cached input | $0.300/Mtok | $0.044/Mtok |
| Price basis | — | DeepSeek bills this model at TWO rates depending on the hour. The tracked figure is the PEAK rate: $1.32 input / $0.044 cached input / $3.96 output per 1M tokens. Off-peak is exactly half: $0.66 / $0.022 / $1.98. Peak hours are 01:00-04:00 and 06:00-10:00 UTC; every other hour is off-peak. Peak is the headline here because DeepSeek defines off-peak as half of peak rather than the other way round, so peak is the list rate, and because a caller who does not schedule around the clock needs the published number to be a ceiling rather than a floor. Halve it for a workload that runs entirely outside those seven hours. Until 16:00 UTC on 2026-08-16 this model billed a single flat $0.435 / $0.003625 / $0.87. |
| Specifications | ||
| Context window | 1M tokens | 1M tokens |
| Parameters | Proprietary | Proprietary |
| Speed (TPS) | — | — |
| Modalities | ||
| Input |
textimage
|
text
|
| Benchmarks source: Vellum LLM Leaderboard | ||
| Avg benchmark score | 83.3 | 73.5 |
| Perf / dollar | 13.9 | 37.1 |
| GPQA Diamond | 93.5 | 90.1 |
| SWE-Bench Verified | 93.4 | 80.6 vendor |
| MCP Atlas | 84.2 | 73.6 |
| BrowseComp | 91.2 | 83.4 |
| Humanity's Last Exam | 56 | 48.2 |
| HumanEval | 76.8 | 80.6 |
| Terminal-Bench 2.1 | 88.3 | — |
| LiveCodeBench | — | 93.5 |
| ARC-AGI 2 | — | 30 |
| AIME 2025 | — | 75 |
| MMMLU | — | 82 |
| BFCL | — | 72 |
| MATH 500 | — | 80 |
| Independent composite scores each on its own scale — not part of the average above | ||
| AA Intelligence Index | 59.7 | 53.2 |
| AA Agentic Index | 54.3 | 49.6 |
| AA-Omniscience Index (−100–100) | 19.7 | 0.8 |
| GDPval-AA v2 | 1668 ELO | 1576.8 ELO |
| LMSYS Chatbot Arena | 1485.3 ELO | 1457.8 ELO |
| Providers | ||
| Available from |
Moonshot — $3.00/$15.00/Mtok
|
DeepSeek — $1.32/$3.96/Mtok
Fireworks — $1.74/$3.48/Mtok
Together — $1.74/$3.48/Mtok
DeepInfra — $1.30/$2.60/Mtok
|
Cost at scale
1M tokens · 50/50 input/output| Volume | Kimi K3 | DeepSeek V4 Pro | Savings |
|---|---|---|---|
| 1M tokens | $6 | $1.98 | $4.02 (67%) |
| 10M tokens | $60 | $19.8 | $40.2 (67%) |
| 100M tokens | $600 | $198 | $402 (67%) |
| 1000M tokens | $6000 | $1980 | $4020 (67%) |
When to pick which
distilled from the pricing and spec data above
Pick Kimi K3 if…
- raw capability matters more than price: 83.3 vs 73.5 average benchmark score
- you want open weights — self-host it, fine-tune it, or exit the API entirely; DeepSeek V4 Pro is closed
Pick DeepSeek V4 Pro if…
- cost dominates: $1.98/Mtok blended vs $6.00 — 67% less on the same 50/50 token mix
- your workload re-reads context (agents, RAG, long chats): cached input costs $0.044/Mtok — 97% off its list input price
- provider flexibility: available from 4 providers, so you can price-shop hosts or fail over
Price movement
recorded changes since we began tracking each model · last checked Sep 2, 2026| Model | Changes | Latest move | Old → new $/M |
|---|---|---|---|
| Kimi K3 | 0 | — no move | Stable since we began tracking — no change recorded |
| DeepSeek V4 Pro | 1 | price rise on Aug 16, 2026 | $0.435/$0.87 → $1.32/$3.96/Mtok |
Only DeepSeek V4 Pro has changed price since we began tracking; Kimi K3 has held steady.
See all recent price moves →
MODELPRICEWATCH.COM · 2026-09-02
Try DeepSeek V4 Pro on DeepSeek
The cheaper option here — DeepSeek V4 Pro costs $1.98/Mtok blended on DeepSeek.
Summary
Kimi K3 by Moonshot costs $3.00/Mtok input and $15.00/Mtok output, with a 1M-token context window. It supports text, image input.
DeepSeek V4 Pro by DeepSeek costs $1.32/Mtok input and $3.96/Mtok output, with a 1M-token context window. It supports text input and is available from 4 providers.
On a blended cost basis, DeepSeek V4 Pro is 67% cheaper than Kimi K3.
On benchmarks, Kimi K3 scores higher (83.3 vs 73.5) on average. In terms of value, DeepSeek V4 Pro has better performance per dollar (37.1 vs 13.9).
Note: Pricing is per million tokens. Actual costs vary with usage patterns, prompt caching, and batch discounts. Always verify against official provider pricing pages.
More comparisons
pairs sharing a model with this page
Claude Fable 5 vs DeepSeek V4 Pro
90% price gap
Claude Fable 5.1 vs DeepSeek V4 Pro
90% price gap
KAT-Coder-Air V2.5 vs DeepSeek V4 Pro
87% price gap
Claude Opus 5 vs DeepSeek V4 Pro
80% price gap
MiniMax-M3 vs DeepSeek V4 Pro
74% price gap
DeepSeek V4 Pro vs Gemini 3.1 Flash-Lite
72% price gap
Kimi K3 vs Kimi K2.7 Code
72% price gap
DeepSeek V4 Pro vs Llama 3.3 70B
68% price gap