Claude Sonnet 5.5 vs Gemini 3.8 Flash
Anthropic's Claude Sonnet 5.5 and Google's Gemini 3.8 Flash, compared on the figures Alevr runs on: list prices, context window, speed and intelligence grades, and the plan that includes each.
Anthropic
Claude Sonnet 5.5
Best speed-to-intelligence balance: near-Opus quality for everyday work.
- Input / output per M
- $2.00 / $10.00
- Context
- 1M
- In Alevr from
- Lite
Gemini 3.8 Flash
Google's flagship multimodal Flash model with hybrid reasoning.
- Input / output per M
- $0.75 / $3.75
- Context
- 1.05M
- In Alevr from
- Lite
Side by side.
| Claude Sonnet 5.5 | Gemini 3.8 Flash | |
|---|---|---|
| Lab | Anthropic | |
| Released | Sep 2026 | Sep 2026 |
| Context window | 1,000,000 tokens | 1,048,576 tokens |
| Input, per M tokens | $2.00 | $0.75 |
| Cached input, per M tokens | $0.10 | $0.08 |
| Output, per M tokens | $10.00 | $3.75 |
| Intelligence grade | 9 / 10Catalog estimate | 9 / 10Catalog estimate |
| Speed grade | 5 / 10Catalog estimate | 8 / 10Catalog estimate |
| Adjustable thinking | Yes | Yes |
| Reads images | Yes | Yes |
| Searches the web in Alevr | Yes | Yes |
| Drives agents and Code | Yes | Yes |
| Included in Alevr from | Lite€10.80 a month incl. VAT | Lite€10.80 a month incl. VAT |
Which to pick.
Each answer below is a comparison of the figures above, nothing more. Costs use a chat turn of 800 input and 500 output tokens.
- For the lowest cost per messageGemini 3.8 Flash
- At list prices, 1,000 requests of a chat turn of 800 input and 500 output tokens cost $6.60 on Claude Sonnet 5.5 and $2.48 on Gemini 3.8 Flash. Gemini 3.8 Flash is about 2.7× cheaper for this workload.
- For the hardest reasoningNo clear difference
- Both carry the same intelligence grade, 9 out of 10. Neither model has an Artificial Analysis measurement in Alevr's latest benchmark sync yet, so both grades are the catalog's estimates and may move when one lands.
- For quick back-and-forthGemini 3.8 Flash
- Gemini 3.8 Flash has the higher speed grade, 8 against 5 out of 10.
- For long documentsNo clear difference
- The context windows are close: 1M and 1.05M tokens.
- On a smaller Alevr planNo clear difference
- Both are included from Lite, €10.80 a month incl. VAT.
Pick Claude Sonnet 5.5
On these figures it does not come out ahead on any measure here. Its lab's own positioning is in the note above; in Alevr you can try both on the same prompt and judge the answers yourself.
Pick Gemini 3.8 Flash
- for the lowest cost per message
- for quick back-and-forth
Use both in Alevr.
Alevr puts Claude, GPT, Gemini and 150+ models in one subscription, hosted in France. Plans start at €10.80 a month incl. VAT.
More comparisons
- Claude Sonnet 5.5 vs GPT-6.1 Sol$2.00 / $10.00 vs $2.00 / $10.00 per M tokens
- Claude Sonnet 5.5 vs GPT-6.1 Sol for coding$2.00 / $10.00 vs $2.00 / $10.00 per M tokens
- Claude Sonnet 5.5 vs GLM-5.3$2.00 / $10.00 vs $1.40 / $4.40 per M tokens
- Claude Sonnet 5.5 vs Muse Spark 1.3$2.00 / $10.00 vs $1.25 / $4.25 per M tokens
- Gemini 3.8 Flash vs GPT-6 Luna$0.75 / $3.75 vs $0.10 / $0.50 per M tokens
- Claude Opus 5.5 vs GPT-6.1 Sol$4.00 / $20.00 vs $2.00 / $10.00 per M tokens
Where these figures come from
Prices are each lab's API list price in US dollars per million tokens, as recorded in the model catalog Alevr bills from. Cached input is the rate a prompt-cache hit is billed at. Alevr subscribers do not pay per token: each plan includes a monthly usage budget.
Intelligence and speed are Alevr's 1 to 10 grades. Where Artificial Analysis has measured a model, the grade is computed from its Intelligence Index and median output speed. A model it has not measured yet carries an estimate from the lab's positioning, and the page says so. The benchmark sync last ran on 22 September 2026.
The figures change when labs change their prices. These pages are rebuilt from the same catalog each time Alevr ships, so they match what the app shows.
OpenRouter, a reseller, lists Gemini 3.8 Flash (google/gemini-3.8-flash) at $0.75 input and $3.75 output per million tokens. Resale rates can differ from the lab's own.