// head_to_head
Claude Haiku 5.5 vs Gemma 4 31B
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.
Claude Haiku 5.5 wins outright — it scores higher and costs less.
Claude Haiku 5.5 leads by 31.1 points overall while listing 25% cheaper per blended million tokens. There is no trade-off to reason about here; the only reason to pick the other is a constraint this table does not show, like an existing contract, a region, or a provider you are already on. The two cost measures disagree here, which is worth knowing: Claude Haiku 5.5 has the lower sticker price, but Gemma 4 31B earns each point of capability for less — $0.0017 against $0.0091 — because per-token rates do not predict how many tokens a model actually spends on a task.
anthropic
Claude Haiku 5.5
- Blended / 1M
- $0.200
- Context
- 1M
- Released
- Oct 7, 2026
- Overall score
- 69.8
Gemma 4 31B
- Blended / 1M
- $0.250
- Context
- 262K
- Released
- Apr 2, 2026
- Overall score
- 38.7
Specs and pricing
| Metric | Claude Haiku 5.5 | Gemma 4 31B |
|---|---|---|
| LiveBench overall Weighted mean of the seven LiveBench category scores, 0–100. Higher is better. | 69.8win | 38.7 |
| Cost per point Measured benchmark spend divided by overall score — dollars per point of capability. | $0.0091 | $0.0017win |
| Blended price / 1M 3:1 input:output mix, the usual shape of production traffic. | $0.200win | $0.250 |
| Input price / 1M | $0.100win | $0.200 |
| Output price / 1M | $0.500 | $0.400win |
| Cached input / 1M Price of an input token served from the prompt cache, where the provider publishes one. | $0.010 | — |
| Context window | 1Mwin | 262K |
| Max output tokens | 128Kwin | 16K |
Benchmarks by category
An overall score averages away the thing you probably care about. These are the seven LiveBench categories scored separately — Claude Haiku 5.5 on top, Gemma 4 31B 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 | Claude Haiku 5.5 | Gemma 4 31B |
|---|---|---|
| Support chatbot 1.2K in / 400 out × 200K requests | $57.52/mo | $80.00/mo |
| RAG assistant 8K in / 600 out × 100K requests | $74.00/mo | $184.00/mo |
| Coding agent 40K in / 4K out × 20K requests | $69.60/mo | $192.00/mo |
| Document extraction 20K in / 1.5K out × 50K requests | $133.00/mo | $230.00/mo |
| Bulk classification 500 in / 20 out × 5M requests | $255.00/mo | $540.00/mo |
Which should you pick?
You are running this at volume
Gemma 4 31B
Lowest measured cost per point of capability at $0.0017 per point — the gap compounds with every request.
Quality matters more than the bill
Claude Haiku 5.5
Highest overall LiveBench score of the two at 69.8.
The workload is coding or agentic work
Claude Haiku 5.5
Leads on long-horizon agentic — 40.0 against 10.7.
You need to fit large documents in one call
Claude Haiku 5.5
Wider context window — 1M against 262K.
Claude Haiku 5.5 vs Gemma 4 31B FAQ
Which is better, Claude Haiku 5.5 or Gemma 4 31B?
Claude Haiku 5.5 wins outright — it scores higher and costs less. Claude Haiku 5.5 leads by 31.1 points overall while listing 25% cheaper per blended million tokens. There is no trade-off to reason about here; the only reason to pick the other is a constraint this table does not show, like an existing contract, a region, or a provider you are already on. The two cost measures disagree here, which is worth knowing: Claude Haiku 5.5 has the lower sticker price, but Gemma 4 31B earns each point of capability for less — $0.0017 against $0.0091 — because per-token rates do not predict how many tokens a model actually spends on a task.
Is Claude Haiku 5.5 cheaper than Gemma 4 31B?
Claude Haiku 5.5 is cheaper. On a 3:1 input:output blend, Claude Haiku 5.5 lists at $0.200 per million tokens and Gemma 4 31B at $0.250 — Claude Haiku 5.5 is 25% cheaper. Input and output are priced separately — Claude Haiku 5.5 charges $0.100 in and $0.500 out, Gemma 4 31B charges $0.200 and $0.400 — so the model that looks cheaper flips depending on how output-heavy your workload is.
Claude Haiku 5.5 vs Gemma 4 31B: which scores higher on benchmarks?
Claude Haiku 5.5 scores 69.8 and Gemma 4 31B scores 38.7 overall on LiveBench, the weighted mean of its seven categories. That is a 31.1-point lead for Claude Haiku 5.5. 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, Claude Haiku 5.5 or Gemma 4 31B?
Gemma 4 31B. Cost per point divides the measured dollars LiveBench spent running the benchmark by the score it earned, so it captures something token pricing misses: a reasoning model can emit many times more tokens than its per-token rate suggests. Claude Haiku 5.5 works out at $0.0091 per point and Gemma 4 31B at $0.0017.
Does Claude Haiku 5.5 or Gemma 4 31B have a bigger context window?
Claude Haiku 5.5 has the larger context window: 1M for Claude Haiku 5.5 against 262K for Gemma 4 31B. 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 5.5 and Gemma 4 31B support prompt caching?
Claude Haiku 5.5 publishes a cached-input rate of $0.010 per million tokens against a full input rate of $0.100. The catalogue lists no separate cached rate for Gemma 4 31B, which means the provider does not price it separately here — not that caching is unavailable.
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 2026-10-25, 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.