Kimi K2 Thinking vs Kimi K2 API pricing
Updated
Kimi K2 Thinking: USD 0.80 input, USD 1.20 output per 1M tokens (GMI Cloud). Kimi K2: USD 0.57 input, USD 2.30 output per 1M tokens (Novita AI). At 3 input tokens per output token, Kimi K2 costs 11% more than Kimi K2 Thinking (USD 1.002 vs USD 0.90 per 1M).
How do Kimi K2 Thinking and Kimi K2 differ on price?
- Cheapest input price: Kimi K2 USD 0.57 against USD 0.60 per 1M tokens, so Kimi K2 Thinking costs 1.05× as much (5% more).
- Cheapest output price: Kimi K2 Thinking USD 1.20 against USD 2.30 per 1M tokens, so Kimi K2 costs 1.92× as much (92% more).
- Among the 201 indexed models, Kimi K2 Thinking ranks 125th cheapest and Kimi K2 132nd on the 3:1 blend.
- The cheaper model changes at 4.8 input tokens per output token.
- Only Kimi K2 Thinking lists a cached-input discount: Kimi K2 Thinking at 2 standard prices of 6, from USD 0.06 (Google Vertex AI, 90% below that provider's input price). Kimi K2's cached tokens bill at the input price.
- Kimi K2 Thinking states up to 262,144 tokens of context, 2× the 131,072 of Kimi K2.
- Kimi K2 Thinking has 6 providers with a standard price, 4 more than Kimi K2; 2 providers sell both (Novita AI and Featherless).
- On one account, Novita AI charges USD 1.075 for Kimi K2 Thinking (1.19× its cheapest price) and USD 1.002 for Kimi K2 (its cheapest price) at 3:1, the lowest combined price among the providers that sell both.
- Shopping around matters more for Kimi K2 Thinking: its dearest standard price is 1.19× its cheapest, against 1.07× for Kimi K2.
- For the one-developer coding agent preset, the gap is USD 9,050 a year in Kimi K2 Thinking's favour (Kimi K2 Thinking USD 478.80 a month, Kimi K2 USD 1,233).
- Per character of input, Kimi K2 Thinking costs 1.05× as much as Kimi K2 in every measured language, because both use the Kimi K3 tokenizer.
Kimi K2 Thinking and Kimi K2 prices side by side
USD per 1M tokens, standard tier. "Cheapest" is the lowest price at 3 input tokens to 1 output token; each line can come from a different provider.
| Price | Kimi K2 Thinking | Kimi K2 |
|---|---|---|
| Model page | Kimi K2 Thinking Moonshot AI · 6 providers | Kimi K2 Instruct Moonshot AI · 2 providers |
| Cheapest standard price, blended 3:1 | USD 0.90 lower USD 0.80 in / USD 1.20 out at GMI Cloud GMI Cloud page checked 2026-11-15 | USD 1.002 USD 0.57 in / USD 2.30 out at Novita AI Novita AI page checked 2026-11-14 |
| Cheapest input | USD 0.60 Amazon Bedrock source | USD 0.57 lower Novita AI source |
| Cheapest cached input | USD 0.06 Google Vertex AI source | No standard price lists a cache price; cached tokens bill at the input price. |
| Cheapest output | USD 1.20 lower GMI Cloud source | USD 2.30 Novita AI source |
| Maker's own API | Moonshot AI publishes no paid standard price on its own API that we found. | Moonshot AI publishes no paid standard price on its own API that we found. |
| Median standard price | USD 0.60 / USD 2.50 input / output, 6 prices | USD 0.585 / USD 2.40 input / output, 2 prices |
| Context window | 262,144 tokens, largest stated; from 32,768 at some providers | 131,072 tokens, largest stated; from 32,768 at some providers |
| Providers with a standard price | 6 | 2 |
- Lower marks the lower price on a line; tie, equal prices. Other tiers and free rows are on the model pages.
Report a wrong price or number (email to info@broadnet.ai, with this page's link filled in).
What do Kimi K2 Thinking and Kimi K2 cost for a real workload?
30 days at each model's cheapest standard price for the workload (presets of the cost calculator; cached input at the cached price where listed).
| Workload | Kimi K2 Thinking | Kimi K2 | Difference |
|---|---|---|---|
| Support chatbot 2,000 requests a day, 1,500 input tokens (60% cached), 300 output | USD 69.84 Google Vertex AI | USD 92.70 Novita AI | USD 22.86 Kimi K2 Thinking lower |
| RAG over documents 1,000 requests a day, 8,000 input tokens (30% cached), 500 output | USD 142.62 Google Vertex AI | USD 171.30 Novita AI | USD 28.68 Kimi K2 Thinking lower |
| Coding agent (one developer) 1,500 requests a day, 40,000 input tokens (85% cached), 2,000 output incl. reasoning | USD 478.80 Google Vertex AI | USD 1,233 Novita AI | USD 754.20 Kimi K2 Thinking lower |
Kimi K2 Thinking is cheaper on all three workloads. The widest gap is on Coding agent (one developer): the dearer model costs 2.58× as much as the cheaper one.
Does the cheaper model change with your input:output ratio?
The cheaper model changes with the mix: below 4.8 input tokens per output token, Kimi K2 Thinking is cheaper; above 4.8, Kimi K2 is cheaper. At the site's 3:1 blend, Kimi K2 Thinking is cheaper. More input per output token, as with retrieval or agents that resend long contexts, favours the lower input price; long answers favour the lower output price.
| Input : output tokens | Kimi K2 Thinking | Kimi K2 |
|---|---|---|
| 1 : 3 (long answers) | USD 1.10 lower GMI Cloud | USD 1.867 Novita AI |
| 1 : 1 | USD 1 lower GMI Cloud | USD 1.435 Novita AI |
| 3 : 1 (the site’s blend) | USD 0.90 lower GMI Cloud | USD 1.002 Novita AI |
| 10 : 1 (retrieval, long prompts) | USD 0.7727 Amazon Bedrock | USD 0.7273 lower Novita AI |
| 30 : 1 (agents, big contexts) | USD 0.6613 Amazon Bedrock | USD 0.6258 lower Novita AI |
USD per 1M tokens at the cheapest standard provider for each mix, before cache discounts. Thinking tokens bill as output.
What do Kimi K2 Thinking and Kimi K2 cost per character in other languages?
Cost to send 1M characters at each model's cheapest non-promotional standard input price, with the Kimi K3 tokenizer for Kimi K2 Thinking (the same one for both, so the gap equals the price gap):
| Content | Kimi K2 Thinking | Kimi K2 |
|---|---|---|
| English | USD 0.130 Amazon Bedrock | USD 0.124 lower Novita AI |
| Arabic MSA | USD 0.288 Amazon Bedrock | USD 0.273 lower Novita AI |
| Egyptian Arabic | USD 0.289 Amazon Bedrock | USD 0.275 lower Novita AI |
| Hindi | USD 0.341 Amazon Bedrock | USD 0.324 lower Novita AI |
| Chinese | USD 0.327 Amazon Bedrock | USD 0.311 lower Novita AI |
| Spanish | USD 0.176 Amazon Bedrock | USD 0.167 lower Novita AI |
| Code | USD 0.156 Amazon Bedrock | USD 0.148 lower Novita AI |
From the real cost by language dataset (measured tokens per character on FLORES-200 and other texts). Your own text will differ.
Check it on your own traffic
- Kimi K2 Thinking API pricing: all 6 providers and other tiers.
- Kimi K2 Instruct API pricing: all 2 providers and other tiers.
- Cost calculator with Kimi K2 Thinking and Kimi K2 preloaded
- Kimi token counter
- Prompt caching calculator
More comparisons: GLM 5.3 vs Kimi K2 · Kimi K2.5 vs Kimi K2 · Kimi K2 0905 vs Kimi K2 · every Kimi K2 Thinking comparison · every Kimi K2 comparison · all 662 price comparisons
Questions
Is Kimi K2 Thinking cheaper than Kimi K2?
Yes. On a blend of 3 input tokens to 1 output token, Kimi K2 Thinking costs USD 0.90 per 1M tokens and Kimi K2 USD 1.002, at each model's cheapest standard price. For input alone, Kimi K2 is cheaper (USD 0.57 against USD 0.60 per 1M); for output alone, Kimi K2 Thinking is (USD 1.20 against USD 2.30). This compares price only, not quality.
What do Kimi K2 Thinking and Kimi K2 cost per 1M tokens?
Kimi K2 Thinking: from USD 0.80 input and USD 1.20 output at GMI Cloud; median of 6 standard prices USD 0.60 / USD 2.50. Kimi K2: from USD 0.57 input and USD 2.30 output at Novita AI; median of 2 standard prices USD 0.585 / USD 2.40. Each price is read from the provider's own pricing page; the source is linked in the table.
Which providers sell both Kimi K2 Thinking and Kimi K2?
2 providers have a standard price for both: Novita AI (USD 1.075 and USD 1.002 at 3:1), Featherless (USD 1.075 and USD 1.075 at 3:1). The first is the lowest combined price. Buying each model from its own cheapest provider costs USD 0.90 and USD 1.002.
Which is cheaper for a coding agent or a chatbot, Kimi K2 Thinking or Kimi K2?
For a coding agent (1,500 requests a day, 40,000 input tokens (85% cached), 2,000 output incl. reasoning): USD 478.80 a month on Kimi K2 Thinking (Google Vertex AI) against USD 1,233 on Kimi K2 (Novita AI), a difference of USD 754.20. For a support chatbot (2,000 requests a day, 1,500 input tokens (60% cached), 300 output): USD 69.84 a month on Kimi K2 Thinking (Google Vertex AI) against USD 92.70 on Kimi K2 (Novita AI), a difference of USD 22.86. Each uses the cheapest standard price for that workload, cached input included. Try your own numbers.
At what input:output ratio does the cheaper of Kimi K2 Thinking and Kimi K2 change?
The cheaper model changes with the mix: below 4.8 input tokens per output token, Kimi K2 Thinking is cheaper; above 4.8, Kimi K2 is cheaper. At the site's 3:1 blend, Kimi K2 Thinking is cheaper. More input per output token, as with retrieval or agents that resend long contexts, favours the lower input price; long answers favour the lower output price.
Do Kimi K2 Thinking and Kimi K2 have cached input pricing?
Kimi K2 Thinking from USD 0.06 per 1M cached input tokens (Google Vertex AI; 2 of 6 standard prices list a discount). Kimi K2: no standard price lists a cached-input discount. Cached tokens are billed at the input price where no cache price is published.
Which has the larger context window, Kimi K2 Thinking or Kimi K2?
Kimi K2 Thinking: up to 262,144 tokens, against 131,072 for Kimi K2, as stated on providers' pricing pages. Some providers offer less than the maximum; check the model pages before you pick the cheapest row for long prompts.