Gemini 3.7 Flash, 3.6 Flash & 3.5 Flash-Lite: Benchmarks, Pricing, Speed

Google just released Gemini 3.7 Flash, three weeks after 3.6 Flash. It scores 40 on the Artificial Analysis Intelligence Index (up from 34 for 3.6 Flash), runs at 288 tok/s, and costs $0.75/$3.75 per million tokens through the end of 2026. That puts it within two points of GPT-5.6 Terra at a fraction of the cost.

The speed of iteration here is unusual. Three Flash releases in under three months (3.5 in May, 3.6 in July, 3.7 in August), each with meaningful intelligence gains. Google credits "developer feedback and algorithmic innovations." Meanwhile, 3.5 Pro still hasn't shipped.

Update (August 13, 2026): Gemini 3.7 Flash launched with a 6-point intelligence jump over 3.6 Flash (40 vs 34 on today's Intelligence Index). Three effort levels (low/medium/high). Same intro pricing as 3.6 Flash ($0.75/$3.75). Available in the API, AI Studio, Antigravity, and Spark. All data below is updated to reflect 3.7 Flash. OpenRouter has it at google/gemini-3.7-flash with an additional 50% discount ($0.375/$1.875) during the intro period.

Update (September 24, 2026): Google has shipped another one. Gemini 3.8 Flash landed on September 2, Google's fourth Flash model in under four months. It scores 41 on the Artificial Analysis Intelligence Index, one point above 3.7 Flash, at the same $0.75/$3.75. At about 297 tok/s, it's the fastest model in AA's top tier, and it scores 42 on AA's Coding Agent Index v1.5. If you're starting today, use 3.8 Flash. AA also rebased its Intelligence Index in September 2026 (now v4.3.2), so every score is lower than at launch: 3.7 Flash went from 56 to 40, and 3.6 Flash from 50 to 34. The comparison tables below use v4.3.2 scores and AA's current prices and speeds. Per-effort scores, launch speeds, and cost per task come from the August launch evaluation and are labeled that way. For live rankings, see the LLM Leaderboard.

Update (September 26, 2026): Gemini 3.8 Flash thinks a lot at high effort, so despite its low list price it costs more per task than Gemini 3.1 Pro, and a full answer takes about 24 seconds. See the LLM leaderboard for real costs per task, and best AI for image analysis for Google's vision picks.

The Flash Family Side by Side

3.7 Flash scores 40 on v4.3.2 at its best setting. It was the strongest Flash-tier model Google had released until 3.8 Flash (41) arrived in September. It edges Grok 4.5 (39), sits two points behind GPT-5.6 Terra (42), and four behind Kimi K3 (44). At $0.75/$3.75, it costs a fraction of any of those competitors. AA now lists all three settings on v4.3.2 (high 39.1, medium 39.6, low 36.9). The speed and cost rows are launch data.

The effort levels give you a useful tradeoff. At launch, 3.7 Flash (low) scored 51 on AA's old v4.1.1 scale and cost $0.16 per task. That matched GPT-5.6 Luna's score at the time while running faster (254 vs 150 tok/s) and costing less per task. The medium level at 53 beat Claude Sonnet 5 on price while nearly matching its launch-era score.

3.7 Flash: What Changed in Three Weeks

The headline: a 6-point intelligence increase with no price increase. 3.7 Flash launched August 13, 2026, exactly 23 days after 3.6 Flash.

Google calls it "our most intelligent workhorse model yet for coding and agents." The improvements come from algorithmic changes, not scale. Same Flash architecture, same pricing tier, but noticeably smarter output. Here's what Artificial Analysis measured:

The cost-per-task drop deserves attention. Despite the same per-token pricing ($0.75/$3.75), 3.7 Flash costs 20% less per task than 3.6 Flash. That means the model is getting better at solving problems with fewer tokens. Smarter and more efficient.

Google specifically mentioned improvements in "debugging, issue resolution, and better first pass code accuracy." Evan Otero from the OpenRouter team confirmed: "Better intelligence for complex workflows. Tons of improvements in debugging, issue resolution, and better first pass code accuracy."

Three effort levels are available, similar to how Claude Opus 5 handles effort:

These per-effort numbers are from AA's August evaluation on the old scale. On that scale, low effort matched Luna's score (51) at a fraction of the speed cost, and high effort competed with models priced 4-8x higher. The low effort level is particularly useful for sub-agent routing where you need "good enough" reasoning at minimal cost.

Pricing runs through the end of 2026 at the introductory $0.75/$3.75. Starting January 1, 2027, it reverts to $1.50/$7.50. On OpenRouter, the current pricing is even lower at $0.375/$1.875 with an additional 50% promotional discount.

Available in the Gemini API, Google AI Studio, Android Studio, Google Antigravity, and through Spark (Google's personal agent for AI Pro and Ultra subscribers).

Where 3.6 Flash Improved Over 3.5 Flash

The headline number: time per task dropped from 2.7 minutes to 1.3 minutes, a 50%+ reduction measured by Artificial Analysis. That's driven by two things working together: 17% fewer output tokens per task and nearly double the output speed (304 tok/s vs 156 tok/s). On some benchmarks, the efficiency gain is much larger. Google says DeepSWE token usage dropped by 65%.

Here's what the benchmark improvements look like:

The coding gains are real. DeepSWE and MLE Bench both saw double-digit improvements. 3.6 Flash "delivers higher precision with fewer unwanted code edits and reduced execution loops," according to Google. In practice, that means less back-and-forth when using it as a coding agent.

Computer use is now a built-in client-side tool through the Gemini API. The 83% OSWorld-Verified score is the highest in Google's comparison table, ahead of GPT-5.6 Luna and Grok 4.5. If you're building browser automation or computer-use agents, 3.6 Flash is worth testing.

The long-context performance jump is the most dramatic. GDM-MRCR v2 at 1M tokens went from under 27% to 54%. That means 3.6 Flash can actually use its full 1M context window without quality collapsing. The previous generation couldn't say that.

Knowledge cutoff moved from January 2025 to March 2026. Fourteen months of fresher training data. That matters for anything involving recent APIs, libraries, or world events.

One caveat: a newer knowledge cutoff doesn't guarantee accurate recall of recent facts. If your workflow depends on the model knowing recent facts, use Search Grounding or RAG rather than trusting the base weights.

3.5 Flash-Lite: The Speed and Cost Story

Flash-Lite isn't trying to compete with frontier models. It's built for the parts of your pipeline where you need fast, cheap, good-enough responses at scale.

Artificial Analysis measured time per task at 0.6 minutes, nearly half the 1.0 minutes of Gemini 3.1 Flash-Lite. Average output tokens per task dropped from 20K to 13K, meaning Flash-Lite is doing the same work with fewer tokens. The cost per task still went up ($0.04 to $0.09) because the per-token pricing increased, but the Intelligence gain is massive: +11 points on the Intelligence Index, with the biggest jumps in agentic evaluations (GDPval-AA went from 642 to 1140, TerminalBench from 31 to 53.6).

The comparison that matters here is against models in the same price bracket:

Flash-Lite is a third of the input cost of GPT-5.4 mini. Less than a third of Claude Haiku 4.5. And it lands within a few points on most benchmarks while leading on some.

What caught my attention: Flash-Lite beats Gemini 3 Flash on SWE-Bench Pro (54.2% vs 49.6%) and OSWorld-Verified (74.0% vs 65.1%). The "Lite" label is misleading. This model is genuinely better than last generation's main Flash at agentic and coding tasks.

Google designed configurable thinking levels for Flash-Lite. You can dial reasoning down to "minimal" for high-volume batch work, or push it to higher thinking for complex sub-agent tasks. That flexibility matters when you're routing different parts of a pipeline through the same model.

Gemini 3.5 Flash Cyber: Google's Security-Focused Model

This one's different from the other two. It's not a general-purpose model you can call through the API.

Gemini 3.5 Flash Cyber is fine-tuned from 3.5 Flash for one purpose: finding and fixing security vulnerabilities. It doesn't run standalone. It works inside CodeMender, Google's code security agent, where multiple Flash Cyber agents coordinate to scan code, validate findings, and produce a combined vulnerability report.

The architecture is multi-agent. Several Flash Cyber instances work in parallel on different aspects of a codebase, each handling discovery or verification. The combined report goes through validation before surfacing results. Google says this reaches "competitive performance at the frontier" on CyberGym.

Why this matters beyond security: Flash Cyber shows that fine-tuning a cheap, fast model for a specific domain and wrapping it in purpose-built agent infrastructure can match frontier performance. The model itself costs less per token than larger alternatives. The agent architecture (CodeMender) is doing the heavy lifting on coordination and accuracy.

The access model is deliberately restricted. Governments and trusted partners only, via a limited pilot. No public API. No timeline for broader access. Google is being cautious, and for good reason. A model optimized for finding vulnerabilities could also help exploit them. The restricted release is the right call.

For most developers, this doesn't change your workflow today. But it signals Google's playbook for specialized AI: take a cheap Flash model, fine-tune it for a domain, wrap it in an agent framework, and deploy it to targeted users. Expect similar specialized models for healthcare, legal, finance, and other verticals. If Kimi K3's unrestricted security capabilities make you nervous, Google's gated approach is the opposite end of that spectrum.

Gemini 3.5 Pro: When Is It Coming Out?

Google announced Gemini 3.5 Pro at I/O 2026 in May. It was supposed to ship in June. Then an intermediate deadline passed. Then July 17. It's missed all of them.

The July 21 blog post said "Gemini 3.5 Pro is currently testing with partners and we plan to make it broadly available as soon as it's ready." The August 13 blog post (for 3.7 Flash) doesn't mention 3.5 Pro at all. That silence says more than any update would.

Bloomberg reported that Google updated the training data in June and got disappointing results on coding. Stories that DeepMind scrapped the model, restarted pretraining after recursive tool-calling and SVG failures, and ran "Rev25" checkpoints trace back to aggregator posts and a single social-media account. ChatForest checked them against Bloomberg and 9to5Google and found the primary reporting doesn't support them.

What Bloomberg did report: engineers are concerned the company risks losing its edge as Anthropic and OpenAI ship stronger models. Google's response: "We're shipping quickly across a wide range of models while keeping them highly cost-effective for customers."

The Flash updates are solid and production-ready. But if you were planning API strategy around Gemini 3.5 Pro benchmarks, don't. No confirmed benchmark numbers exist for it. Build against what you can test today.

And here's the bigger picture. While everyone waits for 3.5 Pro, Google confirmed Gemini 4 pre-training has started. The question developers are starting to ask: does anyone still need 3.5 Pro if 3.7 Flash covers most production use cases and Gemini 4 is the next real leap?

Neither 3.5 Pro nor Gemini 4 is in our September data. The Pro model we track, Gemini 3.1 Pro, scores 30 on the current index at $2/$12. That's below 3.7 Flash (40) and 3.8 Flash (41), at more than twice the price per token.

How 3.7 Flash Compares to the Competition

3.7 Flash changes the competitive picture. At 40 on v4.3.2, it's no longer just a "fast cheap model." It sits within a few points of mid-frontier models at a fraction of the cost. Scores, speeds and prices below are AA's current figures.

3.7 Flash vs GPT-5.6 Terra: Terra scores two points higher (42 vs 40), but 3.7 Flash costs 63% less on input ($0.75 vs $2) and 69% less on output ($3.75 vs $12). Speed isn't close: 288 tok/s vs 83 tok/s. Pick Terra if you need the OpenAI ecosystem, max/ultra reasoning modes, or those two extra points.

3.7 Flash vs Grok 4.5: 3.7 Flash scores slightly higher (40 vs 39) and runs over 4x faster (288 vs 62 tok/s). Grok costs more on both input ($2 vs $0.75) and output ($6 vs $3.75), though its token efficiency (fewer tokens per task) narrows the cost gap. Its successor, Grok 4.7, scores 46 at the same $2/$6. For latency-sensitive work, 3.7 Flash wins. For token-constrained budgets, Grok is competitive.

3.7 Flash vs Kimi K3: K3 still leads on intelligence (44 vs 40) and has open weights. 3.7 Flash is nearly 8x faster (288 vs 37 tok/s) and costs 75% less on output ($3.75 vs $15). For speed-sensitive production workloads, 3.7 Flash wins. For self-hosting, K3 is an option, though MiMo-V2.6-Pro (46) and GLM-5.3 (45) now score higher among open-weight models.

3.7 Flash vs GPT-5.6 Luna: Luna costs less ($0.20/$1.20 vs $0.75/$3.75) but scores 3 points lower (37 vs 40), and 3.7 Flash runs more than twice as fast (288 vs 127 tok/s). Luna wins on pure cost, and GPT-6 Luna now halves it again at $0.10/$0.50 for the same 37. 3.7 Flash wins on intelligence and speed.

3.7 Flash vs 3.6 Flash: Upgrade immediately. +6 points on Intelligence (34 to 40), faster (217 to 288 tok/s on AA's current measurements), 20% cheaper per task at launch ($0.50 to $0.40), same pricing. No reason to stay on 3.6 Flash.

3.8 Flash vs 3.7 Flash: One point higher (41 vs 40), a little faster (297 vs 288 tok/s), same $0.75/$3.75. A small gain, but there's no reason not to take it.

Real-World Cost Comparison

At list prices: 500 tasks/month at 20,000 tokens per task, split evenly between input and output. Thinking tokens aren't included, so use our cost calculator for per-task costs that count them.

3.7 Flash at ~$23/month for 500 tasks is remarkable. It scores 40 on the Intelligence Index while costing less than Sonnet 5, Terra, or Grok 4.5, and it outscores Sonnet 5 and Grok 4.5. Only the Luna models are cheaper, and they score 3 points lower. On the OpenRouter promotional pricing ($0.375/$1.875), the monthly cost drops to ~$11.

At launch, Artificial Analysis measured 3.7 Flash (high) at $0.40 per task and 3.7 Flash (low) at $0.16. On AA's current index runs, Gemini 3.8 Flash costs $1.24 per task at high effort against $1.99 for GPT-5.6 Sol at max.

Best Setup for Agent Workloads

Google clearly built these models for multi-agent architectures. Here's what the pricing and performance data suggests:

Two-model setup within Gemini:

• Master agent: Gemini 3.7 Flash at high effort ($0.75/$3.75). Handles orchestration, complex reasoning, computer use.
• Sub-agents: Gemini 3.7 Flash at low effort ($0.75/$3.75, $0.16/task). Handles routing, data extraction, summarization, bulk processing.

You don't even need two different models. Just switch effort levels. One model, two modes. Gemini 3.8 Flash costs the same and scores one point higher, so it's a drop-in swap for the master agent.

Mixed-provider setup:

• Hard problems: Claude Opus 5.5 (58) or GPT-6 Sol (48) for tasks that need frontier intelligence. Our Opus 5 guide and GPT-5.6 guide cover how they compare with their predecessors.
• Production agents: Gemini 3.8 Flash or 3.7 Flash (high) for speed-sensitive agent loops that need strong intelligence.
• Bulk sub-tasks: Gemini 3.7 Flash (low) or 3.5 Flash-Lite for anything high-volume.

This splits the workload by what each model does best. Sol or Opus handles the 10% of tasks that need frontier intelligence. 3.7 Flash handles the 60% that need speed and solid quality. Flash (low) or Flash-Lite handles the 30% that just need to be fast and cheap.

Use our model selector to find the right fit for your workload.

What Developers Are Saying

The intelligence jump from 34 to 40 (on today's v4.3.2 scale) addressed the main criticism of 3.6 (that it scored below Grok 4.5, Sonnet 5, and even Luna on raw intelligence). At 40, it now sits above all of those.

The rapid iteration is getting attention. Going from 3.5 to 3.6 to 3.7 in under three months suggests Google's Flash team has found a productive optimization loop. Whether they can sustain this pace is an open question, but the results speak for themselves.

The 3.6 Flash criticism was valid and worth revisiting for context. For a similar budget, GLM-5.2 ($1.40/$4.40) scored the same 34 and Grok 4.5 ($2/$6) scored 39. With 3.7 Flash at 40, ahead of GLM-5.2 (34) and Grok 4.5 (39), that complaint no longer applied. GLM-5.3 has since reopened it a little: 45 at $1.40/$4.40.

If your workflow depends on the model knowing recent facts, use Search Grounding or RAG rather than the base weights.

Flash-tier models are designed for speed and cost, not peak intelligence. Comparing them against full-sized frontier models on raw quality is the wrong framing. At 40, though, 3.7 Flash is blurring that line. It's within 4 points of Kimi K3 and 2 of GPT-5.6 Terra. That's not a Flash model you can dismiss as "budget."

Benchmark Gaps Worth Knowing

A few things Google doesn't cover that affect your evaluation.

No SWE-Bench Verified numbers for 3.7 Flash yet. Google reports SWE-Bench Pro but not Verified, which has historically been Anthropic's strongest benchmark. Every lab leads with the benchmarks where they shine.

At launch, Artificial Analysis published full Intelligence Index results across all three effort levels, confirming the intelligence gains. Their independent testing measured 340 tok/s for 3.7 Flash (high), 274 tok/s for medium, and 254 tok/s for low.

The introductory pricing is important to understand. $0.75/$3.75 applies through December 31, 2026. Starting January 1, 2027, the price doubles to $1.50/$7.50. At that point, the cost comparison changes significantly. Plan accordingly if you're building something that depends on the current pricing.

Should You Switch to 3.7 Flash?

Switch from 3.6 Flash immediately. +6 Intelligence points, faster, cheaper per task. Same pricing. This is a free upgrade, and 3.8 Flash adds one more point at the same price.

Switch from 3.5 Flash immediately. 3.7 Flash is better on every dimension. No discussion needed.

Consider switching from GPT-5.6 Terra if you don't need the OpenAI ecosystem. Terra scores 2 points higher (42 vs 40), but 3.7 Flash runs about 3.5x faster and costs 63-69% less.

Consider switching from Grok 4.5 if speed matters more than token efficiency. 3.7 Flash scores slightly higher (40 vs 39) and runs over 4x faster, but Grok uses fewer tokens per task.

Stay on a frontier model if you need intelligence well above 40 on the Index, or you depend on OpenAI/Anthropic-specific tooling. Today that means Claude Opus 5.5 (58) or GPT-6 Sol (48).

Use Flash-Lite if you're running high-volume batch processing, sub-agent tasks, or cost-sensitive pipelines where a score of 22 on the current index is sufficient. At $0.30/$2.50, it's the cheapest option in Google's lineup, though GPT-6 Luna ($0.10/$0.50) now undercuts it and scores higher (37).

FAQ

Is Gemini 3.7 Flash better than 3.6 Flash?

Yes, across the board. 3.7 Flash scores 40 on the Artificial Analysis Intelligence Index (v4.3.2) vs 3.6 Flash's 34. It runs faster (288 vs 217 tok/s), cost 20% less per task at launch ($0.40 vs $0.50), and has the same per-token pricing ($0.75/$3.75 intro). Three effort levels (low/medium/high) give you more granular control. There's no reason to stay on 3.6 Flash, and Gemini 3.8 Flash (41, same price) is now one step newer.

How much does Gemini 3.7 Flash cost?

$0.75 per million input tokens and $3.75 per million output tokens (introductory pricing through December 31, 2026). Starting January 2027, prices double to $1.50/$7.50. On OpenRouter, there's an additional 50% promotional discount bringing it to $0.375/$1.875. Cache hits get a 90% discount.

Is Gemini 3.7 Flash better than GPT-5.6 Luna?

3.7 Flash scores 40 vs Luna's 37 on the Intelligence Index (v4.3.2) and runs more than twice as fast (288 vs 127 tok/s). Luna is cheaper ($0.20/$1.20 vs $0.75/$3.75), and GPT-6 Luna matches its score at $0.10/$0.50. Pick based on whether you need intelligence and speed or the lowest cost.

How fast is Gemini 3.5 Flash-Lite?

350 tokens per second at launch, making it one of the fastest models available. For context, AA currently measures GPT-5.6 Luna at 127 tok/s and Claude Sonnet 5 at 77 tok/s. Flash-Lite is built for workloads where latency matters more than peak intelligence.

What happened to Gemini 3.5 Pro?

It missed its June deadline, an intermediate date, and a July 17 target. The August 13 blog post for 3.7 Flash doesn't mention it at all. Bloomberg reported disappointing coding results after a June retraining. Reports of a full rebuild aren't backed by the primary sources. Don't plan your API strategy around it.

Can I use Gemini 3.5 Flash Cyber?

Not yet through the public API. Flash Cyber is available only to governments and trusted partners through a limited-access pilot via CodeMender, Google's code security agent. There's no timeline for broader availability.

How does Gemini 3.7 Flash compare to Kimi K3?

K3 scores higher on the Intelligence Index (44 vs 40 on v4.3.2) and has open weights. 3.7 Flash is nearly 8x faster (288 vs 37 tok/s) and costs 75% less on output ($3.75 vs $15). For speed-sensitive production workloads, 3.7 Flash wins. For peak coding quality and self-hosting, K3 is the better pick.

Is Gemini 3.5 Flash-Lite good enough for coding?

Yes, for many tasks. It scores 54.2% on SWE-Bench Pro, which beats Gemini 3 Flash (49.6%). It won't replace Sol or Fable 5 for complex debugging. But for code explanation, simple refactoring, test generation, and agent sub-tasks, Flash-Lite handles the work at a fraction of the cost.

When is Gemini 3.5 Pro coming out?

No confirmed date as of our August reporting. Google announced it at I/O 2026 in May and missed every deadline after that. Bloomberg reported that a June retraining gave disappointing coding results. With 3.7 Flash at 40 and 3.8 Flash at 41 on the current index, the gap between Flash and what Pro was expected to deliver keeps shrinking.

Why is Gemini 3.5 Pro delayed?

Bloomberg reported that Google updated the training data in June and got disappointing coding results, and that engineers worry about losing ground to Anthropic and OpenAI. Stories of a scrapped model and a full pretraining restart trace back to aggregator posts and one social-media account, not the primary reporting. Meanwhile Google keeps shipping Flash updates.

What is Gemini 4?

Google confirmed they've started "the most ambitious pre-training run yet" for Gemini 4. No specs, no timeline, no benchmarks. It's a signal that Google's next major model generation is underway, separate from the 3.5 Pro delays. Expect more details later in 2026.

Current scores, prices and speeds from Artificial Analysis Intelligence Index v4.3.2 (September 2026). Per-effort scores, launch speeds and cost per task from the v4.1.1 launch evaluation. Model details from Google's official blog posts for 3.7 Flash and 3.6 Flash. 3.5 Pro delay reporting from Bloomberg, The Verge, and ChatForest. Additional analysis from officechai.com. Updated September 26, 2026. See the LLM Leaderboard for live rankings, Best LLM in 2026 for the full comparison, or the Benchmark Dashboard for side-by-side model analysis.