Model comparison
Mistral Medium 3.5 vs Gemini 3.8 Flash: the 2027 price match
Gemini 3.8 Flash costs half as much as Mistral Medium 3.5 on every rate until December 31, 2026. From January 1, 2027, their list prices match to the cent.
· Prices as of September 28, 2026
Mistral Medium 3.5
Mistral AI · Released May 22, 2026 · Preview
Mistral's coding and agent flagship, a dense open-weight model that replaced Devstral 2 in Mistral's Vibe coding agents.
Mistral Medium 3.5 facts and comparisonsGemini 3.8 Flash
Google · Released September 2, 2026
Google's newest and most capable Flash model, aimed at long-running coding and agent work at Flash prices.
Gemini 3.8 Flash facts and comparisons
The short answer
Gemini 3.8 Flash costs exactly half as much as Mistral Medium 3.5 on every rate today, so the example agentic coding session runs $0.71 against $1.43. That changes on January 1, 2027, when Flash's introductory rates end and its new list prices, $1.50 input, $0.15 cached, and $7.50 output, equal Mistral Medium 3.5's rates. From then on the choice rests on limits and tools: Flash has a 1.05M window and runs in Gemini CLI, Cursor, and GitHub Copilot, while Mistral Medium 3.5 has open weights and a 256K window.
Choose Mistral Medium 3.5 if
- You want weights you can run yourself, which Mistral publishes under a modified MIT license.
- You are planning for 2027, when Flash's list price rises to Mistral Medium 3.5's level and the price advantage goes away.
- You use Mistral's Vibe coding agents, which Medium 3.5 now powers.
- You are drawn to Mistral's description of a frontier-class multimodal model tuned for agentic and coding work.
Choose Gemini 3.8 Flash if
- You want the lower price now: through December 31, 2026, every Flash rate is half of Mistral Medium 3.5's.
- You need more than 256K tokens of context, and Flash's window reaches 1.05M.
- You work in Gemini CLI, or in Cursor, OpenCode, or GitHub Copilot, which list Flash but not Mistral Medium 3.5.
- A zero-cost start helps, and Flash's input, output, and caching are covered by the Gemini API free tier.
Side by side
Specs and prices
| Fact | Mistral Medium 3.5 | Gemini 3.8 Flash |
|---|---|---|
| Maker | Mistral AI | |
| API model id | mistral-medium-3.5 | gemini-3.8-flash |
| Released | May 22, 2026 | September 2, 2026 |
| Status | Preview | Current |
| Context window | 256K tokens | 1.05M tokens |
| Max output | Not published | 65.5K tokens |
| Open weights | Yes | No |
| Input, per 1M tokens | $1.50 | $0.75 |
| Cache hit, per 1M | $0.15 | $0.075 |
| Cache write, per 1M | $1.50 (same as input) | $0.75 (same as input) |
| Output, per 1M tokens | $7.50 | $3.75 |
| Runs in | OpenRouter | Cursor, Gemini CLI, OpenCode, OpenRouter, and GitHub Copilot |
Standard API rates in US dollars, as published by each maker on September 28, 2026. Batch and priority tiers, taxes, and subscription plans are not included. Mistral Medium 3.5: Mistral lists no per-model cache price. Its caching docs bill cached tokens at 10% of input, which is the $0.15 shown. Gemini 3.8 Flash: Introductory rates through December 31, 2026. From January 1, 2027: $1.50 input, $0.15 cached, and $7.50 output per million tokens.
Cost
What the same work costs
The same token counts, priced at each model’s published rates. Cached tokens are billed at each maker’s cache prices, so the session shows what caching is worth on each model.
Real sessions differ: the two models count the same code as different numbers of tokens, and reasoning settings change how much each one writes. Your own history is the real test.
| Workload | Mistral Medium 3.5 | Gemini 3.8 Flash |
|---|---|---|
| Agentic coding session, 100K input, 400K written to cache, 2M read from cache, 50K output | $1.43 | $0.71 |
| Large one-off review, 150K input with no cache hits, 10K output | $0.30 | $0.15 |
| Output-heavy generation, 30K input, 80K output | $0.65 | $0.32 |
| A month of sessions, 110 sessions: 5 a day, 22 working days | $156.75 | $78.38 |
| Where the session’s cost goes | ||
| Cache writes | $0.60 | $0.30 |
| Cache reads | $0.30 | $0.15 |
| Uncached input | $0.15 | $0.08 |
| Output | $0.38 | $0.19 |
- caching saves on the session with Mistral Medium 3.5 (65%)
- $2.70
- caching saves on the session with Gemini 3.8 Flash (66%)
- $1.35
Half the price now, the same price from January 2027
Put the two rate cards side by side and every Mistral number is double. Mistral Medium 3.5 charges $1.50 per million for input, $0.15 for a cache hit, and $7.50 for output, and Gemini 3.8 Flash charges $0.75, $0.075, and $3.75. Neither maker charges a cache-write fee, so writes follow the same 2x ratio.
The workloads follow the same ratio: $0.71 against $1.43 for the agentic session, $0.15 against $0.30 for the large one-off review, $0.32 against $0.65 for the output-heavy generation, and $78.38 against $156.75 for a month of 110 sessions.
Google labels Flash's current prices introductory rates through December 31, 2026. From January 1, 2027 they rise to $1.50 input, $0.15 cached, and $7.50 output, which are Mistral Medium 3.5's rates to the cent. On the published schedules the price gap closes entirely at the new year, and what separates the two afterwards is everything other than price.
What separates them once the price doesn't
Context is the clearest difference. Flash accepts 1.05M tokens and writes up to 65.5K per request. Mistral Medium 3.5 accepts 256K, and Mistral does not publish a maximum output. Neither model's price data lists a long-context tier.
Status differs too. Flash is a current model, released on September 2, 2026, and Google calls it "our most intelligent Flash model, engineered for long-horizon software engineering, autonomous agents, and complex enterprise workflows." Mistral announced Medium 3.5 as a public preview in May 2026 and describes it as "our frontier-class multimodal model optimized for agentic and coding use cases." In Mistral's Vibe coding agents it replaced Devstral 2, which was retired on July 30, 2026.
Then there are the weights. Mistral publishes Medium 3.5's under a modified MIT license and says the dense model runs on as few as four GPUs. Flash's weights are not published, and this post does not estimate hosting costs for either.
Caching rules and where each model runs
Mistral publishes no per-model cache price; the $0.15 in the tables comes from its rule that cached prompt tokens cost 10% of input, the same share Google charges on Flash. Neither maker lists a write fee, so caching saves 65% of Mistral's uncached session and 66% of Flash's. The mechanics differ. Mistral caches prefixes in 64-token blocks, and a prompt_cache_key improves the odds of a hit. Google's implicit caching is on by default without a promised hit, and its explicit caching adds storage at $0.50 to $1 per million tokens per hour on Flash models.
For Gemini API key and Vertex AI users, Gemini CLI, Google's own coding agent, uses Flash for the Flash side of its default auto model. Cursor, OpenCode, OpenRouter, and GitHub Copilot list it as well. Mistral Medium 3.5 is reachable here through OpenRouter, which routes open-weight requests to one of several providers whose prices can differ from Mistral's own API.
EveryToken prices Flash at Google's rates from Gemini CLI, Cursor, or OpenCode history, and prices Mistral Medium 3.5 from OpenRouter's catalog when it runs through OpenRouter, so a session on either shows up as an API-equivalent estimate.
Prompt caching
How each maker bills cached tokens
Mistral AI
Mistral caches prompt prefixes in 64-token blocks. A prompt_cache_key raises the chance of a hit but doesn't guarantee one.
Cached prompt tokens are billed at 10% of the standard input price, and Mistral lists no fee for writing the cache.
Source: Mistral docs: Prompt caching
Implicit caching is on by default for Gemini 2.5 and newer. When a request repeats a prefix Google has cached, the discount is applied automatically, but a hit is not guaranteed.
Explicit caching creates a cache you reference by name, with a guaranteed discount. It adds a storage charge for as long as the cache lives, 1 hour by default: $4.50 per million tokens per hour on Pro models and $0.50 to $1 on Flash models.
On every model compared here, a cache hit costs 10% of the input price. Google publishes no separate cache-write price, so this blog prices written tokens as ordinary input.
Source: Google: Context caching
Your own numbers
See what Mistral Medium 3.5 and Gemini 3.8 Flash really cost you.
everyaitoken reads your OpenRouter, Gemini CLI, Cursor, and OpenCode history on your Mac and prices every request at API rates, with what caching saved or cost. $9 once.
FAQ
Questions
Is Gemini 3.8 Flash half the price of Mistral Medium 3.5?
Through December 31, 2026, yes, on every rate: $0.75 against $1.50 for input, $0.075 against $0.15 for cache hits, and $3.75 against $7.50 for output.
What will Gemini 3.8 Flash cost in 2027?
From January 1, 2027, $1.50 input, $0.15 cached, and $7.50 output per million tokens. Those are the same rates Mistral Medium 3.5 lists today.
Which of the two can read a larger codebase in one request?
Gemini 3.8 Flash, with a 1.05M context window against 256K on Mistral Medium 3.5. Flash caps each response at 65.5K tokens, and Mistral doesn't publish an output cap.
Can I run Mistral Medium 3.5 myself?
Mistral publishes its weights under a modified MIT license and says the dense model can be self-hosted on as few as four GPUs. This post does not estimate what that would cost.
Sources
- Mistral docs: Mistral Medium 3.5
- Mistral: Vibe remote agents and Mistral Medium 3.5
- Mistral docs: Models
- OpenRouter: Mistral Medium 3.5
- Google: Gemini API pricing
- Google docs: Gemini 3.8 Flash
- Google: Gemini models
- Google: Latest Gemini model
- Google DeepMind: Gemini 3.8 Flash model card
- Google: Gemini 3.8 Flash
- Gemini CLI source: model configuration
- Cursor docs: Gemini 3.8 Flash
- OpenRouter: Gemini 3.8 Flash
- OpenCode docs: Zen
- GitHub Docs: Supported AI models in Copilot
- Mistral docs: Prompt caching
- Google: Context caching