// head_to_head

Claude Haiku 4.5 vs Kimi K2.6

Published LiveBench scores across all seven categories, live list pricing, context windows, and the measured cost of a point of capability — for both models, side by side.

OpenRouter + LiveBenchAll comparisonsFull leaderboard

Kimi K2.6 is the cheaper of the two; neither can be ranked on quality here.

Claude Haiku 4.5 does not have a published LiveBench run, so this comparison covers price, context and declared capabilities only. A missing score means "not evaluated", not "worse" — the right way to separate these two is an eval on your own workload.

anthropic

Claude Haiku 4.5

Blended / 1M
$2.00
Context
200K
Released
Oct 15, 2025
Overall score
Not evaluated
reasoningtool callingimage inputfile inputprompt caching

moonshotai

Kimi K2.6

Blended / 1M
$1.71
Context
262K
Released
Apr 20, 2026
Overall score
70.5
reasoningtool callingimage inputprompt caching

Specs and pricing

MetricClaude Haiku 4.5Kimi K2.6
LiveBench overall

Mean of the seven LiveBench category scores, 0–100. Higher is better.

70.5
Cost per point

Measured benchmark spend divided by overall score — dollars per point of capability.

$0.0918
Blended price / 1M

3:1 input:output mix, the usual shape of production traffic.

$2.00$1.71win
Input price / 1M$1.00$0.950
Output price / 1M$5.00$4.00win
Cached input / 1M

Price of an input token served from the prompt cache, where the provider publishes one.

$0.100win$0.160
Context window200K262Kwin
Max output tokens64K236Kwin

Benchmarks by category

An overall score averages away the thing you probably care about. These are the seven LiveBench categories scored separately — Claude Haiku 4.5 on top, Kimi K2.6 below, both out of 100.

Agentic coding
46.9
Coding
78.6
Reasoning
79.4
Mathematics
84.3
Data analysis
65.1
Language
75.1
Instruction following
64.4

What each one costs to run

Per-token prices are hard to feel. These are monthly list costs for both models across five workload shapes, using each provider's published cached-input rate where there is one.

WorkloadClaude Haiku 4.5Kimi K2.6
Support chatbot

1.2K in / 400 out × 200K requests

$575.20/mo$491.12/mo
RAG assistant

8K in / 600 out × 100K requests

$740.00/mo$684.00/mo
Coding agent

40K in / 4K out × 20K requests

$696.00/mo$637.60/mo
Document extraction

20K in / 1.5K out × 50K requests

$1,330/mo$1,211/mo
Bulk classification

500 in / 20 out × 5M requests

$2,550/mo$2,380/mo
Run these two through the cost calculator

Which should you pick?

You need to fit large documents in one call

Kimi K2.6

Wider context window — 262K against 200K.

Claude Haiku 4.5 vs Kimi K2.6 FAQ

Which is better, Claude Haiku 4.5 or Kimi K2.6?

Kimi K2.6 is the cheaper of the two; neither can be ranked on quality here. Claude Haiku 4.5 does not have a published LiveBench run, so this comparison covers price, context and declared capabilities only. A missing score means "not evaluated", not "worse" — the right way to separate these two is an eval on your own workload.

Is Claude Haiku 4.5 cheaper than Kimi K2.6?

Kimi K2.6 is cheaper. On a 3:1 input:output blend, Claude Haiku 4.5 lists at $2.00 per million tokens and Kimi K2.6 at $1.71 — Kimi K2.6 is 17% cheaper. Input and output are priced separately — Claude Haiku 4.5 charges $1.00 in and $5.00 out, Kimi K2.6 charges $0.950 and $4.00 — so the model that looks cheaper flips depending on how output-heavy your workload is.

Does Claude Haiku 4.5 or Kimi K2.6 have a bigger context window?

Kimi K2.6 has the larger context window: 200K for Claude Haiku 4.5 against 262K for Kimi K2.6. Note that a window you can fill is not a window you should fill — retrieval quality usually degrades well before the limit, and you pay for every token you put in it.

Do Claude Haiku 4.5 and Kimi K2.6 support prompt caching?

Both publish a cached-input rate: $0.100 per million for Claude Haiku 4.5 and $0.160 for Kimi K2.6, against full input rates of $1.00 and $0.950. On a workload with a long stable prefix — a system prompt, a tool schema, a retrieved corpus — that changes the economics more than the headline price does.

Related comparisons

How these numbers are produced

  • Price — provider list price from OpenRouter, refreshed every 15 minutes. “Blended” is a 3:1 input:output mix.
  • ScoresLiveBench release 2026-06-25, using their own category map. Each model shows its strongest published run. A blank means “not evaluated”, never “bad”.
  • Cost per point — the measured dollars LiveBench spent on the run, divided by the score it earned.
  • “Win” — awarded only past a threshold: one full point on a benchmark score, 10% on a price, 25% on a context window. Anything tighter reports as a tie, because effort settings alone move a LiveBench score by more than that.

Published benchmarks rank models on someone else's tasks. Before committing, see LLM & agent evaluation for building an eval on your own.