ModelsAtlas
Explore

Google: Gemma 4 31B vs Mistral: Codestral 2508 (batch)

Neither wins outright: Google: Gemma 4 31B is the pick when you need reasoning / thinking mode, Mistral: Codestral 2508 (batch) when cost matters at scale.

Key numbers

SpecGoogle: Gemma 4 31BMistral: Codestral 2508 (batch)
ProviderGoogleMistral AI
Input price (per 1M tokens)$0.20$0.15
Output price (per 1M tokens)$0.40$0.45
Blended price (3:1 input/output)$0.25$0.22
Cheaper than … of all models we track66%68%
Context window262K tokens (~393 pages)256K tokens (~384 pages)
Larger context than … of all models48%45%
Max output per response16K tokens (~12,300 words)205K tokens (~153,600 words)
Input typesimage, text, videotext, file
Output typestexttext
Free variantYesNo
Listed since2026-04-022025-08-01
Advertisement

What it costs for real workloads

Monthly cost at list price for five common workloads. Token counts per unit are stated so you can scale them to your own traffic. A workload that does not fit a model's context window is marked rather than priced.

Workload (per month)Google: Gemma 4 31BMistral: Codestral 2508 (batch)Difference
Customer-support chatbot
50,000 conversations — each a 6-turn conversation (1,500 in / 400 out tokens)
$23.00$20.25Mistral: Codestral 2508 (batch) saves $2.75
RAG search over documents
20,000 questions — each a question answered from ~8 retrieved passages (6,000 in / 500 out tokens)
$28.00$22.50Mistral: Codestral 2508 (batch) saves $5.50
Coding / tool-using agent
2,000 tasks — each a multi-step task with tool calls and file context (60,000 in / 6,000 out tokens)
$28.80$23.40Mistral: Codestral 2508 (batch) saves $5.40
Batch summarisation
10,000 documents — each a 6,000-word report summarised to a page (8,000 in / 600 out tokens)
$18.40$14.70Mistral: Codestral 2508 (batch) saves $3.70
Long-document analysis
500 documents — each a 150-page contract or codebase read in one pass (100,000 in / 2,000 out tokens)
$10.40$7.95Mistral: Codestral 2508 (batch) saves $2.45

Capabilities side by side

CapabilityGoogle: Gemma 4 31BMistral: Codestral 2508 (batch)What it lets you do
Tool / function calling✓ Yes✓ Yescall your APIs and run agent loops
Structured output (JSON schema)✓ Yes✓ Yesreturn JSON that validates against your schema
Reasoning / thinking mode✓ Yes—spend extra tokens thinking before answering hard problems
Image input✓ Yes—read screenshots, charts and scanned pages
Audio input——take speech or audio directly
File / PDF input—✓ Yesaccept documents without your own parsing step
Image output——generate images, not just text
Built-in web search——answer from live web results
Seed (reproducible sampling)✓ Yes✓ Yesrepeat a generation for tests and evals
Log-probabilities✓ Yes—read token confidence for classification and scoring

From each model's published parameters and input/output types. “—” means not advertised, not proven absent.

Price history

DateModelInput / 1MOutput / 1M
2026-10-15Google: Gemma 4 31B$0.20$0.40
2026-09-10Mistral: Codestral 2508 (batch)$0.15$0.45
2026-08-27Google: Gemma 4 31B$0.09$0.34
2026-08-21Google: Gemma 4 31B$0.10$0.34
2026-08-19Google: Gemma 4 31B$0.09$0.34

Which should you choose?

Choose Google: Gemma 4 31B if…

  • you need reasoning / thinking mode — to spend extra tokens thinking before answering hard problems; Mistral: Codestral 2508 (batch) does not advertise it
  • you need image input — to read screenshots, charts and scanned pages; Mistral: Codestral 2508 (batch) does not advertise it
  • you need log-probabilities — to read token confidence for classification and scoring; Mistral: Codestral 2508 (batch) does not advertise it
  • you want to prototype at zero cost — a free variant is listed
  • you want the newer model — released 2026-04-02, 8 months after the other

Choose Mistral: Codestral 2508 (batch) if…

  • cost matters at scale — it is 10% cheaper on a typical 3:1 input/output mix ($20.25 vs $23.00 a month for 50,000 support chats)
  • you need file / PDF input — to accept documents without your own parsing step; Google: Gemma 4 31B does not advertise it
  • you generate long outputs in one go — up to 205K tokens (about 153,600 words) per response

Frequently asked questions

Is Google: Gemma 4 31B better than Mistral: Codestral 2508 (batch)?

Neither wins outright: Google: Gemma 4 31B is the pick when you need reasoning / thinking mode, Mistral: Codestral 2508 (batch) when cost matters at scale. In short: Google: Gemma 4 31B when you need reasoning / thinking mode, you need image input, you need log-probabilities, you want to prototype at zero cost, you want the newer model; Mistral: Codestral 2508 (batch) when cost matters at scale, you need file / PDF input, you generate long outputs in one go.

Which is cheaper, Google: Gemma 4 31B or Mistral: Codestral 2508 (batch)?

Mistral: Codestral 2508 (batch) is cheaper: $0.22 vs $0.25 per 1M tokens at a typical 3:1 input/output mix, 10% less. For 50,000 support conversations a month that is $20.25 against $23.00.

How much do Google: Gemma 4 31B and Mistral: Codestral 2508 (batch) cost per million tokens?

Google: Gemma 4 31B costs $0.20 per 1M input tokens and $0.40 per 1M output tokens; Mistral: Codestral 2508 (batch) costs $0.15 input and $0.45 output. Output tokens are billed separately, so long answers shift the gap.

Which has the bigger context window, Google: Gemma 4 31B or Mistral: Codestral 2508 (batch)?

Google: Gemma 4 31B: 262K tokens, about 393 pages, against 256K (about 384 pages) for Mistral: Codestral 2508 (batch).

Do Google: Gemma 4 31B and Mistral: Codestral 2508 (batch) support function calling and JSON output?

Tool calling — Google: Gemma 4 31B: yes, Mistral: Codestral 2508 (batch): yes. JSON-schema structured output — Google: Gemma 4 31B: yes, Mistral: Codestral 2508 (batch): yes.

Can Google: Gemma 4 31B or Mistral: Codestral 2508 (batch) read images?

Only Google: Gemma 4 31B accepts image input; Mistral: Codestral 2508 (batch) is text-only for input.

Is Google: Gemma 4 31B or Mistral: Codestral 2508 (batch) free to use?

Google: Gemma 4 31B has a free variant (google/gemma-4-31b-it:free). Free variants usually carry tighter rate limits.

What is the maximum output length of Google: Gemma 4 31B and Mistral: Codestral 2508 (batch)?

Google: Gemma 4 31B can write up to 16K tokens (about 12,300 words) per response; Mistral: Codestral 2508 (batch) up to 205K (about 153,600 words).

When were Google: Gemma 4 31B and Mistral: Codestral 2508 (batch) released?

Google: Gemma 4 31B was listed on 2026-04-02; Mistral: Codestral 2508 (batch) on 2025-08-01.

Has the price of Google: Gemma 4 31B or Mistral: Codestral 2508 (batch) changed?

Google: Gemma 4 31B: input price raised 55% ($0.09 → $0.20/1M), output price raised 15% ($0.34 → $0.40/1M) on 2026-10-15 (4 price changes since 2026-08-06). Mistral: Codestral 2508 (batch): input price cut 50% ($0.30 → $0.15/1M), output price cut 50% ($0.90 → $0.45/1M) on 2026-09-10.

About the two models

Google

Google: Gemma 4 31B

Gemma 4 31B Instruct is Google DeepMind's 30.7B dense multimodal model supporting text and image input with text output. Features a 256K token context window, configurable thinking/reasoning mode, native function...

Full Google: Gemma 4 31B specs →

Compare Google: Gemma 4 31B with other models

vs AI21: Jamba Large 1.7vs AionLabs: Aion 3.5vs AionLabs: Aion 3.5 Minivs AionLabs: Aion-1.0

Compare Mistral: Codestral 2508 (batch) with other models

vs AI21: Jamba Large 1.7vs AionLabs: Aion 3.5vs AionLabs: Aion 3.5 Minivs AionLabs: Aion-1.0
Advertisement