// head_to_head
Kimi K3 vs GPT-5.6 Terra
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.
GPT-5.6 Terra is the better model, at roughly the same price.
GPT-5.6 Terra leads by 2.2 points overall and the two list within about 10% of each other, so the cheaper-but-weaker trade-off does not apply. Price parity plus a score gap usually makes this an easy call.
moonshotai
Kimi K3
- Blended / 1M
- $4.14
- Context
- 1.0M
- Released
- Jul 16, 2026
- Overall score
- 65.0
openai
GPT-5.6 Terra
- Blended / 1M
- $4.50
- Context
- 1.1M
- Released
- Jul 9, 2026
- Overall score
- 67.2
Specs and pricing
| Metric | Kimi K3 | GPT-5.6 Terra |
|---|---|---|
| LiveBench overall Weighted mean of the seven LiveBench category scores, 0–100. Higher is better. | 65.0 | 67.2win |
| Cost per point Measured benchmark spend divided by overall score — dollars per point of capability. | $0.0217 | $0.0209 |
| Blended price / 1M 3:1 input:output mix, the usual shape of production traffic. | $4.14 | $4.50 |
| Input price / 1M | $0.192win | $2.00 |
| Output price / 1M | $16.00 | $12.00win |
| Cached input / 1M Price of an input token served from the prompt cache, where the provider publishes one. | $0.182 | $0.200 |
| Context window | 1.0M | 1.1M |
| Max output tokens | 944Kwin | 128K |
Benchmarks by category
An overall score averages away the thing you probably care about. These are the seven LiveBench categories scored separately — Kimi K3 on top, GPT-5.6 Terra below, both out of 100.
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.
| Workload | Kimi K3 | GPT-5.6 Terra |
|---|---|---|
| Support chatbot 1.2K in / 400 out × 200K requests | $1,325/mo | $1,310/mo |
| RAG assistant 8K in / 600 out × 100K requests | $1,110/mo | $1,600/mo |
| Coding agent 40K in / 4K out × 20K requests | $1,428/mo | $1,552/mo |
| Document extraction 20K in / 1.5K out × 50K requests | $1,392/mo | $2,810/mo |
| Bulk classification 500 in / 20 out × 5M requests | $2,075/mo | $5,300/mo |
Which should you pick?
Quality matters more than the bill
GPT-5.6 Terra
Highest overall LiveBench score of the two at 67.2.
The workload is coding or agentic work
Kimi K3
Leads on long-horizon agentic — 36.0 against 25.9.
Kimi K3 vs GPT-5.6 Terra FAQ
Which is better, Kimi K3 or GPT-5.6 Terra?
GPT-5.6 Terra is the better model, at roughly the same price. GPT-5.6 Terra leads by 2.2 points overall and the two list within about 10% of each other, so the cheaper-but-weaker trade-off does not apply. Price parity plus a score gap usually makes this an easy call.
Is Kimi K3 cheaper than GPT-5.6 Terra?
They cost about the same. Both land near $4.14 per million tokens on a 3:1 input:output blend, so price is unlikely to be the deciding factor between them.
Kimi K3 vs GPT-5.6 Terra: which scores higher on benchmarks?
Kimi K3 scores 65.0 and GPT-5.6 Terra scores 67.2 overall on LiveBench, the weighted mean of its seven categories. That is a 2.2-point lead for GPT-5.6 Terra. Category scores differ from the overall figure — a model can lead on reasoning and trail on coding, which the per-category table above breaks out.
Which gives better value for money, Kimi K3 or GPT-5.6 Terra?
They are close. Kimi K3 costs $0.0217 per point of overall capability and GPT-5.6 Terra costs $0.0209, a difference small enough that workload shape will matter more than the rate.
Does Kimi K3 or GPT-5.6 Terra have a bigger context window?
They are effectively the same — 1.0M for Kimi K3 and 1.1M for GPT-5.6 Terra.
Do Kimi K3 and GPT-5.6 Terra support prompt caching?
Both publish a cached-input rate: $0.182 per million for Kimi K3 and $0.200 for GPT-5.6 Terra, against full input rates of $0.192 and $2.00. 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.
- Scores — LiveBench release 2026-11-06, 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.