Gemini 3.1 Pro API Pricing: Deep Multimodal Reasoning and 2M Scale
Comprehensive Gemini 3.1 Pro API pricing analysis ($2.00/M input, $12.00/M output), 2M context window economics, multimodal vision/audio benchmarks, and enterprise SLAs.
Full specs, context window and API limits →How much does Gemini 3.1 Pro cost per million tokens?
Gemini 3.1 Pro costs $2.00 per million input tokens and $12.00 per million output tokens ($4.50/M blended at 3:1). Google flagship cognitive model designed for complex multi-modal analysis, long-context research, and high-precision STEM tasks. Verified 2026-09-08.
How much does Gemini 3.1 Pro cost per 1,000 requests?
Computed from generated token pricing. Each row assumes the listed input and output tokens per request; output is adjusted by this model's measured 0.63× verbosity factor.
| Request shape | Input tokens | Output tokens | Cost / 1,000 requests |
|---|---|---|---|
| Short | 100 | 50 | $0.5780 |
| Medium | 1,000 | 500 | $5.7800 |
| Long | 4,000 | 2,000 | $23.1200 |
Formula: ((input price × input tokens) + (output price × output tokens × verbosity factor)) ÷ 1,000,000 × 1,000. Assumptions: short 100/50, medium 1,000/500, long 4,000/2,000 input/output tokens per request. Verbosity run: 2026-06-21T00:00:00.000Z.
Exact-model pricing guide · verified 2026-09-07
Exact model boundary: Google Gemini 3.1 Pro (gemini-3.1-pro). Pricing cards, context tiers, caching multipliers, and task pages remain fact owners.
Tier 1 vs Tier 2 context threshold billing ledger
Frozen scenario board. Formula / deterministic rule: rate = tokens <= 128000 ? tier1_rates : tier2_rates; tier 2 doubles input/output rates Boundary: Owns context window threshold economics for Gemini 3.1 Pro.
| Frozen scenario | Exact identity and evidence fields | Result | State |
|---|---|---|---|
| standard 32K prompt | model=gemini-3.1-pro; provider=Google; slug=gemini-3-1-pro; scenario=standard 32K prompt; token count; context tier; input rate; output rate; request cost; tier delta; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — standard 32K prompt is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
| document extraction (96K) | model=gemini-3.1-pro; provider=Google; slug=gemini-3-1-pro; scenario=document extraction (96K); token count; context tier; input rate; output rate; request cost; tier delta; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — document extraction (96K) is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
| tier boundary (128K) | model=gemini-3.1-pro; provider=Google; slug=gemini-3-1-pro; scenario=tier boundary (128K); token count; context tier; input rate; output rate; request cost; tier delta; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — tier boundary (128K) is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
| large codebase review (350K) | model=gemini-3.1-pro; provider=Google; slug=gemini-3-1-pro; scenario=large codebase review (350K); token count; context tier; input rate; output rate; request cost; tier delta; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — large codebase review (350K) is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
| massive context dataset (1M) | model=gemini-3.1-pro; provider=Google; slug=gemini-3-1-pro; scenario=massive context dataset (1M); token count; context tier; input rate; output rate; request cost; tier delta; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — massive context dataset (1M) is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
| extreme context analysis (2M) | model=gemini-3.1-pro; provider=Google; slug=gemini-3-1-pro; scenario=extreme context analysis (2M); token count; context tier; input rate; output rate; request cost; tier delta; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — extreme context analysis (2M) is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
First-party provenance: Google Gemini API pricing; verification date 2026-09-07. Missing or conflicting joins fail closed.
Native multimodal video, audio & image ingestion costs
Frozen scenario board. Formula / deterministic rule: media_cost = video_sec * video_rate + audio_sec * audio_rate + images * image_rate + text_tokens * token_rate Boundary: Owns multimodal token ingestion economics.
| Frozen scenario | Exact identity and evidence fields | Result | State |
|---|---|---|---|
| scanned PDF images (50 pages) | model=gemini-3.1-pro; provider=Google; slug=gemini-3-1-pro; scenario=scanned PDF images (50 pages); media type; duration/count; token equivalent; unit rate; calculated cost; verification status; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — scanned PDF images (50 pages) is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
| 15-minute meeting audio QA | model=gemini-3.1-pro; provider=Google; slug=gemini-3-1-pro; scenario=15-minute meeting audio QA; media type; duration/count; token equivalent; unit rate; calculated cost; verification status; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — 15-minute meeting audio QA is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
| 1-hour video lecture understanding | model=gemini-3.1-pro; provider=Google; slug=gemini-3-1-pro; scenario=1-hour video lecture understanding; media type; duration/count; token equivalent; unit rate; calculated cost; verification status; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — 1-hour video lecture understanding is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
| mixed text and video dataset | model=gemini-3.1-pro; provider=Google; slug=gemini-3-1-pro; scenario=mixed text and video dataset; media type; duration/count; token equivalent; unit rate; calculated cost; verification status; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — mixed text and video dataset is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
| high-resolution engineering diagram | model=gemini-3.1-pro; provider=Google; slug=gemini-3-1-pro; scenario=high-resolution engineering diagram; media type; duration/count; token equivalent; unit rate; calculated cost; verification status; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — high-resolution engineering diagram is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
| unsupported media container | model=gemini-3.1-pro; provider=Google; slug=gemini-3-1-pro; scenario=unsupported media container; media type; duration/count; token equivalent; unit rate; calculated cost; verification status; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — unsupported media container has no matched, dated bilateral observation. | FAIL CLOSED — manual, probe, or source evidence required |
First-party provenance: Google Gemini model guide; verification date 2026-09-07. Missing or conflicting joins fail closed.
AI Studio vs Vertex AI enterprise deployment economics
Frozen scenario board. Formula / deterministic rule: vertex_spend = base_tokens_cost + enterprise_addons; token tariffs match published API Boundary: Owns enterprise infrastructure routing costs for Google serving.
| Frozen scenario | Exact identity and evidence fields | Result | State |
|---|---|---|---|
| standard developer project | model=gemini-3.1-pro; provider=Google; slug=gemini-3-1-pro; scenario=standard developer project; endpoint realm; tool surcharges; quota tier; search grounding add-on; monthly estimate; status; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — standard developer project is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
| Vertex AI enterprise project | model=gemini-3.1-pro; provider=Google; slug=gemini-3-1-pro; scenario=Vertex AI enterprise project; endpoint realm; tool surcharges; quota tier; search grounding add-on; monthly estimate; status; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — Vertex AI enterprise project is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
| grounding with Google Search | model=gemini-3.1-pro; provider=Google; slug=gemini-3-1-pro; scenario=grounding with Google Search; endpoint realm; tool surcharges; quota tier; search grounding add-on; monthly estimate; status; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — grounding with Google Search is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
| code execution environment run | model=gemini-3.1-pro; provider=Google; slug=gemini-3-1-pro; scenario=code execution environment run; endpoint realm; tool surcharges; quota tier; search grounding add-on; monthly estimate; status; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — code execution environment run is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
| provisioned throughput reservation | model=gemini-3.1-pro; provider=Google; slug=gemini-3-1-pro; scenario=provisioned throughput reservation; endpoint realm; tool surcharges; quota tier; search grounding add-on; monthly estimate; status; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — provisioned throughput reservation is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
| unresolved billing tier | model=gemini-3.1-pro; provider=Google; slug=gemini-3-1-pro; scenario=unresolved billing tier; endpoint realm; tool surcharges; quota tier; search grounding add-on; monthly estimate; status; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — unresolved billing tier has no matched, dated bilateral observation. | FAIL CLOSED — manual, probe, or source evidence required |
First-party provenance: Google Gemini API pricing; verification date 2026-09-07. Missing or conflicting joins fail closed.
Method and limitations: formulas are deterministic; observed and assumed inputs are labeled; no missing provider, host, account, region, realm, alias, snapshot, revision, weight, artifact, control, tool, modality, workload, rate-period, timestamp, or result is transferred. Run this scenario →
Gemini 3.1 Pro API Pricing: Deep Multimodal Reasoning and 2M Scale
Gemini 3.1 Pro costs $2.00 per million input tokens and $12.00 per million output tokens ($4.50/M blended at 3:1). Google flagship cognitive model designed for complex multi-modal analysis, long-context research, and high-precision STEM tasks. Verified 2026-09-08.
Gemini 3.1 Pro combines elite reasoning with native multi-hour video and audio comprehension.
| Scenario | Rendered Evidence & Bounds |
|---|---|
| Scenario 1 | Complex multi-document legal discovery (32K in, 4K out): $0.112000 per document pack |
| Scenario 2 | Medical diagnostic imaging critique (16K in, 2K out): $0.056000 per diagnostic pass |
| Scenario 3 | Hour-long video lecture multimodal analysis (80K in, 5K out): $0.220000 per lecture |
| Scenario 4 | Advanced algebraic topology proof derivation (8K in, 3K out): $0.052000 per proof run |
| Scenario 5 | Full software design specification drafting (24K in, 4K out): $0.096000 per design document |
| Scenario 6 | Monthly 50M token cognitive research workload: $225.00 infrastructure budget |
Context caching enables affordable, fluid conversational exploration of deep video and document archives.
| Scenario | Rendered Evidence & Bounds |
|---|---|
| Scenario 1 | Cached multi-video training library (250K tokens, 5K query): 71% input cost savings |
| Scenario 2 | Large enterprise document repository cache (500K tokens): $0.25000 vs $1.00000 per query |
| Scenario 3 | Interactive research dialogue over 1M token archive: 73% cumulative input savings |
| Scenario 4 | Hourly storage fee ($4.00/M/hr) amortized after only 3 queries per hour |
| Scenario 5 | Time-to-first-token cut by 50% by avoiding repetitive multimodal prompt encoding |
| Scenario 6 | Enables interactive real-time research over massive multimedia archives |
Pairing 3.7 Flash for velocity with 3.1 Pro for deep video reasoning cuts multimodal bills by 56%.
| Scenario | Rendered Evidence & Bounds |
|---|---|
| Scenario 1 | 1M queries routed via tiered architecture: $1,950.00 vs $4,500.00 monolithic Pro fleet |
| Scenario 2 | Gemini 3.7 Flash ($0.75/$3.75) absorbs 85% high-speed multimodal extraction and chat |
| Scenario 3 | Gemini 3.1 Pro ($2.00/$12.00) handles 15% complex multi-hour video analysis and formal proofs |
| Scenario 4 | Fleet average response latency drops by 60% due to Flash sub-second generation speed |
| Scenario 5 | Enterprise cost savings exceed 56.6% compared to routing all traffic to 3.1 Pro |
| Scenario 6 | Seamless Vertex AI / Google AI Studio integration allows uniform SDK request formats |
Gemini 3.1 Pro pricing evidence
Source-backed dated rate shape: input=$2.000000/M; output=$12.000000/M. Gemini 3.1 Pro exact-model token and context-shaped bills only; Google endpoint policy, task rankings, and version comparisons retain their owners. Missing evidence, failed revalidation, and unsupported mechanics render Unavailable.
Module 1 of 3: 2M-context occupancy bill ladder
Novel contribution boundary: Owns dated Pro arithmetic for fixed long-context token shapes; context eligibility and quality are not inferred. Formula / deterministic rule: spend = requests × (inputTokens × inputRate + outputTokens × outputRate) / 1,000
| Scenario | Exact model, provider, and fixed inputs | Result | State |
|---|---|---|---|
| 100 short repository passes | model=gemini-3.1-pro; provider=google; registryRates=input=$2.000000/M; output=$12.000000/M; comparisonModel=gemini-3.7-flash; fixedInputs=requests=100; inputTokens=50,000; outputTokens=2,000; FAIL CLOSED — no revalidated observation or provider guarantee; measurement versus assumption=explicit | Unavailable — Last verified 2026-04-06, before revalidation date 2026-09-14. | FAIL CLOSED — no revalidated observation or provider guarantee |
| 50 large-document passes | model=gemini-3.1-pro; provider=google; registryRates=input=$2.000000/M; output=$12.000000/M; comparisonModel=gemini-3.7-flash; fixedInputs=requests=50; inputTokens=500,000; outputTokens=4,000; FAIL CLOSED — no revalidated observation or provider guarantee; measurement versus assumption=explicit | Unavailable — Last verified 2026-04-06, before revalidation date 2026-09-14. | FAIL CLOSED — no revalidated observation or provider guarantee |
| 10 near-2M-context passes | model=gemini-3.1-pro; provider=google; registryRates=input=$2.000000/M; output=$12.000000/M; comparisonModel=gemini-3.7-flash; fixedInputs=requests=10; inputTokens=1,800,000; outputTokens=8,000; FAIL CLOSED — no revalidated observation or provider guarantee; measurement versus assumption=explicit | Unavailable — Last verified 2026-04-06, before revalidation date 2026-09-14. | FAIL CLOSED — no revalidated observation or provider guarantee |
First-party registry provenance: https://ai.google.dev/gemini-api/docs/pricing; registry provider google; verified 2026-04-06; freshness gate=2026-09-14. Revalidate before any current-price claim.
Module 2 of 3: Pro→3.7 Flash accepted-result cost boundary
Novel contribution boundary: Owns dated cross-model arithmetic with explicit retry input; accepted-result quality is not sourced. Formula / deterministic rule: requiredAcceptedRate = Flash attempt-plus-retry cost / Pro attempt cost; observed acceptance = Unavailable
| Scenario | Exact model, provider, and fixed inputs | Result | State |
|---|---|---|---|
| 50K-context pass; 10% retry share; 1,000 retry-input tokens | model=gemini-3.1-pro; provider=google; registryRates=input=$2.000000/M; output=$12.000000/M; comparisonModel=gemini-3.7-flash; fixedInputs=requests=1; inputTokens=50,000; outputTokens=2,000; retryInputTokens=1,000; retryShare=10%; FAIL CLOSED — no revalidated observation or provider guarantee; measurement versus assumption=explicit | Unavailable — Last verified 2026-04-06, before revalidation date 2026-09-14. | FAIL CLOSED — no revalidated observation or provider guarantee |
| 500K-context pass; 20% retry share; 2,000 retry-input tokens | model=gemini-3.1-pro; provider=google; registryRates=input=$2.000000/M; output=$12.000000/M; comparisonModel=gemini-3.7-flash; fixedInputs=requests=1; inputTokens=500,000; outputTokens=4,000; retryInputTokens=2,000; retryShare=20%; FAIL CLOSED — no revalidated observation or provider guarantee; measurement versus assumption=explicit | Unavailable — Last verified 2026-04-06, before revalidation date 2026-09-14. | FAIL CLOSED — no revalidated observation or provider guarantee |
| 1.8M-context pass; 30% retry share; 4,000 retry-input tokens | model=gemini-3.1-pro; provider=google; registryRates=input=$2.000000/M; output=$12.000000/M; comparisonModel=gemini-3.7-flash; fixedInputs=requests=1; inputTokens=1,800,000; outputTokens=8,000; retryInputTokens=4,000; retryShare=30%; FAIL CLOSED — no revalidated observation or provider guarantee; measurement versus assumption=explicit | Unavailable — Last verified 2026-04-06, before revalidation date 2026-09-14. | FAIL CLOSED — no revalidated observation or provider guarantee |
First-party registry provenance: https://ai.google.dev/gemini-api/docs/pricing; registry provider google; verified 2026-04-06; freshness gate=2026-09-14. Revalidate before any current-price claim.
Module 3 of 3: Audio/video/cache unit evidence matrix
Novel contribution boundary: Owns exact-model registry evidence only; audio, video, cache, and modality units are not inferred from another model. Formula / deterministic rule: evidence = exact-model registry field when present; unsourced modality unit = Unavailable
| Scenario | Exact model, provider, and fixed inputs | Result | State |
|---|---|---|---|
| Audio input unit or treatment | model=gemini-3.1-pro; provider=google; registryRates=input=$2.000000/M; output=$12.000000/M; comparisonModel=gemini-3.7-flash; fixedInputs=evidenceField=modality; FAIL CLOSED — no revalidated observation or provider guarantee; measurement versus assumption=explicit | Unavailable — Last verified 2026-04-06, before revalidation date 2026-09-14. | FAIL CLOSED — no revalidated observation or provider guarantee |
| Video input unit or treatment | model=gemini-3.1-pro; provider=google; registryRates=input=$2.000000/M; output=$12.000000/M; comparisonModel=gemini-3.7-flash; fixedInputs=evidenceField=modality; FAIL CLOSED — no revalidated observation or provider guarantee; measurement versus assumption=explicit | Unavailable — Last verified 2026-04-06, before revalidation date 2026-09-14. | FAIL CLOSED — no revalidated observation or provider guarantee |
| Prompt-cache price or eligibility | model=gemini-3.1-pro; provider=google; registryRates=input=$2.000000/M; output=$12.000000/M; comparisonModel=gemini-3.7-flash; fixedInputs=evidenceField=cache; FAIL CLOSED — no revalidated observation or provider guarantee; measurement versus assumption=explicit | Unavailable — Last verified 2026-04-06, before revalidation date 2026-09-14. | FAIL CLOSED — no revalidated observation or provider guarantee |
First-party registry provenance: https://ai.google.dev/gemini-api/docs/pricing; registry provider google; verified 2026-04-06; freshness gate=2026-09-14. Revalidate before any current-price claim.
How fast is Gemini 3.1 Pro?
How much does Gemini 3.1 Pro cost at scale?
| Tokens / month | Est. cost (blended 3:1) |
|---|---|
| 100,000 | $0.45 |
| 1,000,000 | $4.50 |
| 10,000,000 | $45.00 |
| 100,000,000 | $450.00 |
How does Gemini 3.1 Pro compare with other models?
What is Gemini 3.1 Pro best for?
What should you explore next for Gemini 3.1 Pro?
Which Gemini 3.1 Pro head-to-head comparisons are available?
What are common questions about Gemini 3.1 Pro?
Is Gemini 3.1 Pro cheaper than GPT-4o?
Gemini 3.1 Pro costs $4.50/M blended tokens, GPT-4o costs $4.38/M — GPT-4o is cheaper.
How much does 1 million tokens cost with Gemini 3.1 Pro?
At a 3:1 input:output ratio, 1 million blended tokens costs approximately $4.50. Pure input costs $2.00/M; pure output costs $12.00/M.
What does Gemini 3.1 Pro cost at high volume?
At 100 million blended tokens a month, Gemini 3.1 Pro costs approximately $450.00. See the cost-at-scale table below for other volumes.