Mistral · Mistral Large · Frontier
Mistral Large 2 pricing and limits
Mistral's large general-purpose model, available on la Plateforme and major clouds. Output is only three times input, and there is no discounted cache-read rate in the list price.
| Line | Tokens | Rate / 1M | Cost |
|---|---|---|---|
| Input, uncached | 12,000 | $2.00 | $0.0240 |
| Output | 1,200 | $6.00 | $0.00720 |
- Input / 1M
- $2.00
- Output / 1M
- $6.00
- Cached / 1M
- n/a
- Cache write / 1M
- n/a
- Blended 3:1
- $3.00
- Context
- 128K
What drives Mistral Large 2 cost
Without a cache-read discount, every input token is billed at the full rate. Agents that resend long tool schemas on each step pay for them every time, so prompt compression and trimming tool lists per step are the main input levers.
Function calling and JSON mode are supported; schema-validation retries are a quiet cost driver when strict output formats are required.
Cost of a typical call
Take a agent step with 12,000 input tokens and 1,200 output tokens. At Mistral Large 2 list prices ($2.00 input, $6.00 output per 1M) one call costs $0.0312, and 100,000 calls a month come to $3,120. The list price has no discounted cache-read rate, so every input token is billed at the full rate on every call.
Blended at a 3:1 input to output ratio, Mistral Large 2 costs $3.00 per 1M tokens. Routing the same call to the fallback in the policy below, Mistral Medium 3, costs $0.00720, which is what a tenant pays you in COGS once they cross the hard cap and PALM SOUQ downgrades instead of blocking.
Cost comparison
Mistral Large 2 against three alternatives
The same agent step priced on Mistral Medium 3, GPT-4.1, Claude Sonnet 4.5. Move the sliders to match your traffic; models without a cache-read rate bill every input token in full.
Benchmark quality per task before switching. A cheaper model that needs more steps or longer outputs can cost more per finished task than its rate suggests, which is why PALM SOUQ reports cost per feature, not only per token.
Mistral Large 2 this guide
$0.0312 per call
$3,120 / mo
baseline
Mistral Medium 3
$0.00720 per call
$720 / mo
-77% vs Mistral Large 2
GPT-4.1
$0.0336 per call
$3,360 / mo
+8% vs Mistral Large 2
Claude Sonnet 4.5
$0.0540 per call
$5,400 / mo
+73% vs Mistral Large 2
List prices from the PALM SOUQ LLM Price Index, computed per call as uncached input + cache reads + output. Reasoning tokens count as output on reasoning models.
Rate limits by usage tier
La Plateforme applies rate limits per workspace by tier, covering requests per second and tokens per minute and per month. Limits increase on paid tiers and by request.
Exact limits vary by account tier and change over time, so we do not reprint them here. Check the limits page for your own account in the Mistral documentation at docs.mistral.ai. PALM SOUQ enforces your own per-tenant limits in front of the provider's, so a single customer cannot consume the shared allowance that every other tenant depends on.
Circuit-breaker best practices
Because input is never discounted, a max step cap has a direct effect on cost: each extra step pays full price for the whole accumulated context.
Use schema-validation retry caps and route repeated failures to a human review queue rather than retrying indefinitely.
When a breaker trips, the gateway answers in under 50 ms with HTTP 429 and X-PalmSouq-Reason: circuit_breaker_triggered, before the request reaches Mistral. Your agent framework sees a normal rate-limit response and stops, and the audit log records which heuristic fired and for which tenant.
# palmsouq.policy.yaml
policy: mistral-large-business
match:
model: mistral-large
tier: business # from X-PalmSouq-Tier
budgets:
monthly_usd: 300 # per tenant
soft_cap_pct: 80 # webhook to your app
on_hard_cap: fallback # or: block (HTTP 429)
fallback:
- mistral-medium-3
breakers:
tool_thrashing: { identical_calls: 3, window_steps: 8 }
token_velocity: { max_tokens_per_s: 50000, window_s: 10 }
semantic_echo: { repeats: 3 }
max_steps: 15- budgetsPer-tenant monthly dollar cap with an 80% soft-cap webhook, keyed by X-PalmSouq-Tenant.
- fallbackAt the hard cap, requests reroute to Mistral Medium 3 ($0.80 blended per 1M) instead of failing.
- breakersAll four loop heuristics on, evaluated in the request path with under 5 ms added latency.
Where it fits
European data residency requirements, multilingual workloads and teams standardizing on Mistral across sizes.
Provider documentation: docs.mistral.ai
Mistral Large 2 pricing questions
How much does Mistral Large 2 cost per million tokens?
At list price Mistral Large 2 costs $2.00 per 1M input tokens and $6.00 per 1M output tokens. Prices were last reviewed in the PALM SOUQ LLM Price Index and are for the standard tier and shortest context band.
Does Mistral Large 2 support prompt caching?
The Mistral list price for Mistral Large 2 has no discounted cache-read rate, so PALM SOUQ meters every input token at $2.00 per 1M. Shortening repeated context and caching full responses in your own application are the main savings levers.
What are the Mistral Large 2 rate limits?
La Plateforme applies rate limits per workspace by tier, covering requests per second and tokens per minute and per month. Limits increase on paid tiers and by request. For your account's exact numbers, see the Mistral documentation at docs.mistral.ai.
How do I stop runaway Mistral Large 2 agent loops?
Because input is never discounted, a max step cap has a direct effect on cost: each extra step pays full price for the whole accumulated context. The sample policy on this page enables all four PALM SOUQ heuristics with a max step cap of 15 and falls back to Mistral Medium 3 at the hard cap.
Is there a batch discount for Mistral Large 2?
50% off through the Mistral batch API. Batch traffic suits evaluations, backfills and nightly enrichment jobs that do not need a real-time answer.
Govern Mistral Large 2 spend per customer.
Point your Mistral client at gateway.palmsouqltd.com, pass X-PalmSouq-Tenant, and every Mistral Large 2 token is attributed, capped and protected by circuit breakers.
14-day free production trial · 100k requests included · no card required