
Gemini 3.8 Flash vs 3.1 Pro: Start With Flash
By Nathan Cole
MyClaw Editorial
Compare in MyClaw
Compare Gemini 3.8 Flash and 3.1 Pro on your own work
Run the same coding task in both models to compare speed, cost, tool use, and verified results before choosing your default.
AI Takeaway
- Which should you pick first? Gemini 3.8 Flash for new coding and agent jobs. Keep Gemini 3.1 Pro when a custom-tools harness or a hard eval already wins on it.
- Is Flash cheaper than Pro? Yes on the rate card. Through 31 December 2026 Flash is $0.75 / $3.75 per 1M tokens, then $1.50 / $7.50. Pro is $2 / $12 up to 200K input and $4 / $18 above. 3.8 can spend more tokens than 3.7 because it works harder. It still undercuts Pro's lowest paid band.
- Same window? Yes: 1,048,576 input / 65,536 output.
- Should you drop 3.1 Pro? Not if the job is already tuned to
gemini-3.1-pro-preview-customtools, or Pro still closes a ticket Flash fails. Measure a closed job, not the sticker.
The Matchup Is Current Flash Against a Preview Pro
Gemini 3.8 Flash shipped 2 September 2026 as stable gemini-3.8-flash. Gemini 3.1 Pro shipped 19 February 2026 as preview gemini-3.1-pro-preview. That is six months and three Flash drops. Pro is no longer the obvious first pick.
If a roundup still treats 3.1 Pro as the Gemini you start on, the Flash side has moved. Swap the model ID. Do not fall back to a 3.7 comparison unless that is the ID you still run. The Gemini 3.8 Flash vs 3.7 Flash page stays inside Flash. This page is Flash versus Pro.
3.1 Pro remains a preview model for complex problem-solving and precise tool use. Gemini 3.8 Flash Cyber is a Fairwind Program variant for trusted defenders. Public work stays on gemini-3.8-flash or gemini-3.1-pro-preview.
What 3.8 Flash Is For
3.8 Flash is the step up inside the Flash line, not a Pro-priced model. Official docs call it the most intelligent Flash model, built for long-horizon software engineering, autonomous agents, and complex enterprise workflows, at Flash speed and cost.
Thinking is low, medium, and high. The default is medium. minimal returns an error. Computer use is preview. File search is supported. Google says 3.8 Flash works harder: extra reasoning steps, iterative tool calls, more tokens at higher effort.
On MyClaw there is no 3.8 model page yet. The live Flash listing is still Gemini 3.5 Flash. Run the IDs the provider actually exposes, and write down any substitution.

What 3.1 Pro Still Is For
Keep 3.1 Pro when a workflow already depends on it. The live Gemini 3.1 Pro page is the MyClaw listing. The API ID is still gemini-3.1-pro-preview. Official docs have not marked it generally available.
Google ships a second endpoint, gemini-3.1-pro-preview-customtools, for harnesses that mix bash with custom tools such as repository search or file inspection. File search on the Gemini API is limited to AI Studio. Computer use is not listed as supported. Thinking runs at low, medium, and high, with a default of high. Pin the same level on both models before you compare bills.
A stable Flash endpoint still needs monitoring and a fallback. A six-month preview needs those plus an eye on the ID itself.
Specs at a Glance
Official Gemini API model pages plus the 2 September 2026 launch post, checked 3 September 2026. Regional, batch, Flex, and contract rates can differ.
| Gemini 3.8 Flash | Gemini 3.1 Pro Preview | |
|---|---|---|
| Role | Current Flash workhorse | Preview Pro fallback |
| API ID | gemini-3.8-flash | gemini-3.1-pro-preview |
| Status | Stable (Sep 2026) | Preview (Feb 2026) |
| Context / max output | 1,048,576 / 65,536 | 1,048,576 / 65,536 |
| Intro / standard API price | $0.75 / $3.75 through 31 Dec 2026; $1.50 / $7.50 from 1 Jan 2027 | $2 / $12 up to 200K input; $4 / $18 above |
| Thinking | low, medium, high (default medium) | low, medium, high (default high) |
| Notable extra | File search; computer use preview | gemini-3.1-pro-preview-customtools |
| First use | New coding and agent loops | Proven custom-tool or hard-eval jobs |
| Caveat | Same sticker as 3.7, often more tokens | Preview, 200K price step |
The Rate Card Favors Flash. The Bill Still Needs a Closed Ticket
Through 31 December 2026, Flash input is 62.5% cheaper than Pro's lowest paid band, and output is 68.75% cheaper. Flash also skips the 200K step that doubles Pro input to $4 and lifts output to $18. From 1 January 2027 Flash moves to $1.50 / $7.50. That is still below $2 / $12 if Pro pricing stays unchanged.

Cheaper rates are not a free pass. Extra reasoning steps and extra tool turns show up as output tokens. A verbose agent loop can narrow the gap even when the rate card does not move. Count tokens, retries, and leftover cleanup on one closed ticket. That total is the comparison. The list price is only the rate.
Independent cost-per-task currently favors that reading. Artificial Analysis puts Gemini 3.8 Flash (high) around 59 on the Intelligence Index versus about 48 for 3.1 Pro Preview, and prices those 3.8 runs at about $0.58 per task at roughly 300 tokens per second. Treat those dollars as a directional signal, not your production invoice. Your harness will move the bill more than a leaderboard average.
The older Gemini 3.7 Flash vs 3.1 Pro piece is still useful on how to run a side-by-side. Keep the files and the pass/fail rule. Swap in the 3.8 ID.
Coding and Agents: Upgrade the Default, Keep a Pro Fallback
Use 3.8 Flash for the work you would have sent to Pro six months ago: multi-hour coding, terminal agents, and multi-step research that has to stay coherent. Google's launch pitch concentrates there: DeepSWE v1.1 long-horizon engineering, finance and legal agent benches, and 54.9% on HLE-Verified. Those are vendor scores against a mixed field, not a head-to-head with 3.1 Pro.
3.1 Pro's published cards are from February 2026, including 80.6% on SWE-Bench Verified and 68.5% on Terminal-Bench 2.0. Different dates, different harnesses. They are a reason to test 3.8 on your hardest Flash-priced jobs, not to pick a winner from two press tables.
Keep 3.1 Pro when the job is already passing on the custom-tools endpoint, or when a high-thinking 3.8 pass fails a check Pro still hits. Everyday patches can look closer than the long-horizon table. If a bounded unit-tested change already closes on 3.8, stay on 3.8. Do not migrate a stable Pro loop because the Flash version number moved.
If the split is a coding-agent workflow, freeze the repo, the tools, and the acceptance tests before you swap IDs. The model is only the reasoning layer. The loop is what you are measuring.
Run the Same Job on Flash, Then on Pro

The model stops if the laptop sleeps, the terminal dies, or the files live in another chat. OpenClaw is the runtime: channels, tools, memory, and scheduled jobs. MyClaw hosts that runtime so the workspace stays up while you swap Flash and Pro IDs.
Step 1: Freeze one real coding or agent job
Pick a single outcome: a failing test, a bounded refactor, a sourced research memo, or a scheduled monitor. Prefer a job you already ran on 3.1 Pro or on 3.7 Flash. Write the pass/fail rule before either model starts.
Step 2: Run 3.8 Flash, then 3.1 Pro at a comparable thinking setting
Keep files, tools, permissions, and instructions identical. First pass is 3.8 Flash at medium or high. Second pass is 3.1 Pro at the same thinking level when the ID is actually available; otherwise stay on Flash and record the substitution. Track tokens, retries, leftover cleanup, and whether the ticket closed.
Step 3: Keep the hosted workspace so you can switch
Leave logs, diffs, notes, and artifacts in the same hosted workspace. The next run should start from the work, not a blank chat. MyClaw pricing covers the always-on host. Model tokens still bill separately.
Pick by Job Shape, Then Keep a Fallback
| If you need… | Start with… |
|---|---|
| New long-horizon coding or agent loops | Gemini 3.8 Flash |
| A harness already tuned to custom tools | Gemini 3.1 Pro Preview |
| Lower token overhead on 3.8 | 3.8 Flash at a lower thinking level |
| Overnight / always-on execution | A hosted OpenClaw workspace, then swap the model |
Keep a cheaper backup and three to five jobs you re-run when prices or model IDs move. Start a new Gemini 3.8 Flash vs 3.1 Pro job on Flash. Keep 3.1 Pro when the closed ticket says so.
Compare in MyClaw
Compare Gemini 3.8 Flash and 3.1 Pro on your own work
Run the same coding task in both models to compare speed, cost, tool use, and verified results before choosing your default.