Gemini 4 Hydrogen vs Claude Sonnet 5.5
Gemini 4 Hydrogen ($1.50/$7.50) vs Claude Sonnet 5.5 ($2/$10): which wins agentic cost per task?
01 // VERDICT
The short answer
| At a glance | Gemini 4 Hydrogen | Claude Sonnet 5.5 |
|---|---|---|
| List priceinput / output per 1M tokens | $1.5 / $7.5 (better) | $2 / $10 |
| Benchmark task costArtificial Analysis, per Intelligence-Index task | $3.16 (better) | $10.02 |
| Intelligence IndexAA v5.0.1, higher is better | 54 | 55 (better) |
Highlighted values mark the lower rate or stronger measured score in the same documented run. · Gemini pricing snapshot (verified 2026-11-19)
EXECUTIVE SUMMARY
Gemini 4 Hydrogen undercuts Claude Sonnet 5.5 across every line item: 25% lower base pricing ($1.50/$7.50 vs $2.00/$10.00 per MTok, both flat across 1M context), zero cache-write fee ($0 implicit vs $2.50 5m write on Sonnet), lower cache-read rate ($0.075 vs $0.10), plus Free tier and 50%-off Flex serving. In Artificial Analysis Index v5.0.1, Hydrogen (Xhigh) achieves nearly the identical top Intelligence Index score (54 vs Sonnet Max 55) at 3.2x lower realized cost ($3.16 vs $10.02/task) and over twice the speed (14.9 min at 172 tok/s vs 32.3 min at 147 tok/s); at default High effort (51 vs Sonnet High 42), Hydrogen wins by +9 Index points at nearly identical task cost ($1.87 vs $1.64). Hydrogen leads AutomationBench-AA (76.1% vs 71.8%), AA-Omniscience (46 vs 30), HLE-Diamond (58.8% vs 53.3%), and DeepMind's SWE-sweep (13.5% vs 3.1%), while Sonnet 5.5 maintains clear wins on Terminal-Bench (63.6% vs 53.5% AA; 63.9% vs 57.0% Google TB 4.1), AA-Briefcase (1815 vs 1568 Elo), and OSWorld 2.1 (83.9% vs 77.6%).
Choose Gemini 4 Hydrogen
Cost-efficient reasoning, web/API automation (AutomationBench-AA 76.1% vs 71.8%), factual reliability (AA-Omniscience 46 vs 30), long outputs (>128K up to 1M via background decode), cache-write-heavy prompts ($0 write vs $2.50), or Flex/Free-tier usage.
Choose Claude Sonnet 5.5
Shell/terminal recovery (Terminal-Bench 4.0 63.6% vs 53.5% AA), office deliverables (AA-Briefcase 1815 vs 1568 Elo), GUI computer use (OSWorld 2.1 83.9% vs 77.6%), or Anthropic prompt-cache TTL guarantees.
TIE-BREAKER RULE
Either — decide by evidence you generate
Compare effort-matched tiers cleanly: Hydrogen default High (Index 51, $1.87/task) crushes Sonnet High (Index 42, $1.64/task), while Sonnet Max (55, $10.02) stays justified only for shell-heavy Terminal-Bench or Briefcase office work.
02 // KEY FACTS
Key facts
- Both use flat 1M-context pricing (no >272K cliff): Hydrogen lists 25% cheaper at $1.50/$7.50 ($0.075 implicit cached, $0 write) vs Sonnet 5.5 $2.00/$10.00 ($0.10 cached, $2.50 5m write).
- On AA v5.0.1 top settings, Hydrogen Xhigh reaches Index 54 at $3.16/task (14.9 min, 172 tok/s) vs Sonnet 5.5 Max Index 55 at $10.02/task (32.3 min, 147 tok/s) — 3.2x cheaper and 2.2x faster for -1 Index point.
- Effort-matched comparison strongly favors Hydrogen: at 'xhigh' Hydrogen leads 54 vs 50 ($3.16 vs $4.09); at 'high' Hydrogen leads 51 vs 42 ($1.87 vs $1.64).
- Sub-benchmark split: Hydrogen leads AutomationBench-AA (76.1% vs 71.8%), AA-Omniscience (46 vs 30), HLE-Diamond (58.8% vs 53.3%), CritPt v2 (66.0% vs 59.0%), and SWE-sweep (13.5% vs 3.1%).
- Sonnet 5.5 holds clean edges on Terminal-Bench 4.0 (63.6% vs 53.5% AA; 63.9% vs 57.0% Google cheating-adjusted TB 4.1), AA-Briefcase v1.1 (1815 vs 1568 Elo), and OSWorld 2.1 (83.9% vs 77.6%).
- Serving & limits: Hydrogen supports Free tier, Batch/Flex (-50% = $0.75/$3.75), Priority (1.8x), and up to 1,048,576 background output tokens; Sonnet 5.5 supports Standard & Batch (-50% = $1/$5) with 128K max output.
03 // TIER PRICING
Official pricing, every tier
USD per 1M tokens from vendor pricing pages. 'Not published' means exactly that — we don't extrapolate.
| Service tier | Gemini 4 Hydrogen | Claude Sonnet 5.5 |
|---|---|---|
| Standarddefault on-demand serving | in $1.5 · cached $0.075 · out $7.5 implicit cache only (no write fee) | in $2 · cached $0.10 · out $10 cache write $2.5 |
| Batchasync jobs, 50% off | in $0.75 · cached $0.0375 · out $3.75 implicit cache only (no write fee) | in $1 · cached $0.05 · out $5 cache write $1.25 · derived |
| Flexslower/cheaper serving, 50% off | in $0.75 · cached $0.0375 · out $3.75 implicit cache only (no write fee) | Not published |
| Long-context band | Gemini 4 Hydrogen | Claude Sonnet 5.5 |
|---|---|---|
| Long-context ratesapplies per request above the model threshold | Flat pricing same rates at any input size (up to context window) | Flat pricing same rates at any input size (up to context window) |
USD per 1M tokens. Every input token is billed in exactly one bucket — uncached, cached read, or cache write; a write replaces the input rate for those tokens (official rule), so write rates are per tier as listed. Anthropic rows bill the 5-minute cache-write TTL ($5/MTok on Opus 5.5, $10/MTok on Fable 5.2); 1-hour writes ($8 and $16) are also published, and cache multipliers stack with Batch and Fast-mode pricing. · Anthropic pricing (verified 2026-11-19) · These Gemini models use implicit caching. Enter cache hits as cached input and set explicit cache writes to zero. · Google Gemini pricing (verified 2026-11-19)
Gemini service tiers
04 // COST SIMULATOR
Run this comparison on your own numbers
Preloaded with a workload typical for this comparison. Change anything — results recompute locally.
Estimated monthly ledger
browser-localToken-based workload estimate only; it does not measure quality or cost per successful task.
| Model | Per request | Monthly |
|---|---|---|
| Gemini 4 Hydrogencheapest | $0.0486 | $1,701.00 |
| Claude Sonnet 5.5 | $0.0648 | $2,268.00 |
Monthly spread between the cheapest and priciest selected model: $567.00.
Breakdown: Gemini 4 Hydrogen · standard tier · standard context
- Fresh input (uncached): 12,000 tok × $1.5/M$0.018
- Cached input (reads): 8,000 tok × $0.075/M$0.0006
- Cache writes: 0 tok · rate unavailable$0.00
- Output (incl. reasoning tokens): 4,000 tok × $7.5/M$0.03
- Per request$0.0486
Rates: official pricing · verified 2026-11-19
- Per-token arithmetic only: it does not model quality differences or cross-vendor output-token-count differences for the same task.
- Reasoning tokens are billed as output tokens at the chosen model's output rate.
- Long-context band is decided per request by total input tokens vs the provider threshold, not by monthly volume.
- Cache accounting uses the official mutually-exclusive buckets: each input token is billed once, as uncached input, cached read, OR cache write — writes replace the input rate for those tokens, never add to it.
- Claude rows bill the 5-minute-TTL cache write; the published 1-hour TTL ($8/MTok standard) is not modeled (no TTL input).
- Rows flagged 'derived' are computed from a stated official rule (e.g. Anthropic batch -50%), not read verbatim from a price table.
- Model and tier combinations without a published rate appear as Unsupported.
05 // BENCHMARK LOG
What the evidence actually says
Independent and vendor-run evals, each with publisher, runner, and harness. Numbers from different runs are never merged.
| Model | Score | Effort | Harness | Evidence |
|---|---|---|---|---|
| GDPval-AAv2.1 | ||||
| Claude Sonnet 5.5claude-sonnet-5-5 | 1840 elo | — | Anthropic Haiku 5.5 announcement evals (GDPval-AA benchmark by Artificial Analysis) | Vendor-reported |
| AA-Briefcase (v1.1)v1.1 | ||||
| Claude Sonnet 5.5claude-sonnet-5-5 | 1824 elo | — | Anthropic Haiku 5.5 announcement evals | Vendor-reported |
| OSWorld (2.1 offline subset)2.1 offline subset | ||||
| Claude Sonnet 5.5claude-sonnet-5-5 | 83.9 % | — | Anthropic Haiku 5.5 announcement evals | Vendor-reported |
| Terminal-Bench (4.0)4.0 | ||||
| Claude Sonnet 5.5claude-sonnet-5-5 | 70.6 % | — | Anthropic Haiku 5.5 announcement evals | Vendor-reported |
| FrontierCode1.1 (Main) | ||||
| Claude Sonnet 5.5claude-sonnet-5-5 | 52.1 % | xhigh | Anthropic Haiku 5.5 announcement evals | Vendor-reported |
| Chartographyunspecified (no tools) | ||||
| Claude Sonnet 5.5claude-sonnet-5-5 | 61.6 % | — | Anthropic Haiku 5.5 announcement evals | Vendor-reported |
| AA Intelligence Indexv5.0.1 | ||||
| Gemini 4 Hydrogengemini-4-hydrogen | 54 index-points | config 'Gemini 4 Hydrogen (xhigh)' per AA | Artificial Analysis Intelligence Index v5.0.1 (independent, same harness across all models) | Third-party |
| Claude Sonnet 5.5claude-sonnet-5-5 | 55 index-points | config 'Claude Sonnet 5.5 (max)' per AA | Artificial Analysis Intelligence Index v5.0.1 (independent, same harness across all models) | Third-party |
| AA cost per Intelligence-Index taskv5.0.1 | ||||
| Gemini 4 Hydrogengemini-4-hydrogen | 3.16 usd-per-task | config 'Gemini 4 Hydrogen (xhigh)' per AA | Artificial Analysis Intelligence Index v5.0.1 (independent, same harness across all models) | Third-party |
| Claude Sonnet 5.5claude-sonnet-5-5 | 10.02 usd-per-task | config 'Claude Sonnet 5.5 (max)' per AA | Artificial Analysis Intelligence Index v5.0.1 (independent, same harness across all models) | Third-party |
| AutomationBench-AAAA v5.0.1 | ||||
| Gemini 4 Hydrogengemini-4-hydrogen | 76.1 % | config 'Gemini 4 Hydrogen (xhigh)' per AA | Artificial Analysis Intelligence Index v5.0.1 (independent, same harness across all models) | Third-party |
| Claude Sonnet 5.5claude-sonnet-5-5 | 71.8 % | config 'Claude Sonnet 5.5 (max)' per AA | Artificial Analysis Intelligence Index v5.0.1 (independent, same harness across all models) | Third-party |
| Terminal-Bench (4.0 (AA v5.0.1))4.0 (AA v5.0.1) | ||||
| Gemini 4 Hydrogengemini-4-hydrogen | 53.5 % | config 'Gemini 4 Hydrogen (xhigh)' per AA | Artificial Analysis Intelligence Index v5.0.1 (independent, same harness across all models) | Third-party |
| Claude Sonnet 5.5claude-sonnet-5-5 | 63.6 % | config 'Claude Sonnet 5.5 (max)' per AA | Artificial Analysis Intelligence Index v5.0.1 (independent, same harness across all models) | Third-party |
| FrontierCode-AA2.0 (AA v5.0.1) | ||||
| Gemini 4 Hydrogengemini-4-hydrogen | 39 % | config 'Gemini 4 Hydrogen (xhigh)' per AA | Artificial Analysis Intelligence Index v5.0.1 (independent, same harness across all models) | Third-party |
| Claude Sonnet 5.5claude-sonnet-5-5 | 41.7 % | config 'Claude Sonnet 5.5 (max)' per AA | Artificial Analysis Intelligence Index v5.0.1 (independent, same harness across all models) | Third-party |
| AA-Briefcase (v1.1 (AA v5.0.1))v1.1 (AA v5.0.1) | ||||
| Gemini 4 Hydrogengemini-4-hydrogen | 1568 elo | config 'Gemini 4 Hydrogen (xhigh)' per AA | Artificial Analysis Intelligence Index v5.0.1 (independent, same harness across all models) | Third-party |
| Claude Sonnet 5.5claude-sonnet-5-5 | 1815 elo | config 'Claude Sonnet 5.5 (max)' per AA | Artificial Analysis Intelligence Index v5.0.1 (independent, same harness across all models) | Third-party |
| AA-Omnisciencev1.1 (AA v5.0.1) | ||||
| Gemini 4 Hydrogengemini-4-hydrogen | 46 index-points | config 'Gemini 4 Hydrogen (xhigh)' per AA | Artificial Analysis Intelligence Index v5.0.1 (independent, same harness across all models) | Third-party |
| Claude Sonnet 5.5claude-sonnet-5-5 | 30 index-points | config 'Claude Sonnet 5.5 (max)' per AA | Artificial Analysis Intelligence Index v5.0.1 (independent, same harness across all models) | Third-party |
| HLE-DiamondAA v5.0.1 | ||||
| Gemini 4 Hydrogengemini-4-hydrogen | 58.8 % | config 'Gemini 4 Hydrogen (xhigh)' per AA | Artificial Analysis Intelligence Index v5.0.1 (independent, same harness across all models) | Third-party |
| Claude Sonnet 5.5claude-sonnet-5-5 | 53.3 % | config 'Claude Sonnet 5.5 (max)' per AA | Artificial Analysis Intelligence Index v5.0.1 (independent, same harness across all models) | Third-party |
| CritPt v2 (Hard Subset)AA v5.0.1 | ||||
| Gemini 4 Hydrogengemini-4-hydrogen | 66 % | config 'Gemini 4 Hydrogen (xhigh)' per AA | Artificial Analysis Intelligence Index v5.0.1 (independent, same harness across all models) | Third-party |
| Claude Sonnet 5.5claude-sonnet-5-5 | 59 % | config 'Claude Sonnet 5.5 (max)' per AA | Artificial Analysis Intelligence Index v5.0.1 (independent, same harness across all models) | Third-party |
| Terminal-Bench-Science (0.1 (AA v5.0.1))0.1 (AA v5.0.1) | ||||
| Gemini 4 Hydrogengemini-4-hydrogen | 49.5 % | config 'Gemini 4 Hydrogen (xhigh)' per AA | Artificial Analysis Intelligence Index v5.0.1 (independent, same harness across all models) | Third-party |
| Claude Sonnet 5.5claude-sonnet-5-5 | 53.3 % | config 'Claude Sonnet 5.5 (max)' per AA | Artificial Analysis Intelligence Index v5.0.1 (independent, same harness across all models) | Third-party |
| GDP.pdfAA v5.0.1 | ||||
| Gemini 4 Hydrogengemini-4-hydrogen | 27.6 % | config 'Gemini 4 Hydrogen (xhigh)' per AA | Artificial Analysis Intelligence Index v5.0.1 (independent, same harness across all models) | Third-party |
| Claude Sonnet 5.5claude-sonnet-5-5 | 25.8 % | config 'Claude Sonnet 5.5 (max)' per AA | Artificial Analysis Intelligence Index v5.0.1 (independent, same harness across all models) | Third-party |
| SWE-sweepGoogle Nov 2026 card | ||||
| Gemini 4 Hydrogengemini-4-hydrogen | 13.5 % | — | Google DeepMind Gemini 4 Hydrogen Model Card evaluation table (2026-11-12) | Vendor-reported |
| Claude Sonnet 5.5claude-sonnet-5-5 | 3.1 % | — | Google DeepMind Gemini 4 Hydrogen Model Card evaluation table (2026-11-12) | Vendor-reported |
| Terminal-Bench (4.1 (cheating-adjusted, Google Nov 2026 card))4.1 (cheating-adjusted, Google Nov 2026 card) | ||||
| Gemini 4 Hydrogengemini-4-hydrogen | 57 % | — | Google DeepMind Gemini 4 Hydrogen Model Card evaluation table (2026-11-12) | Vendor-reported |
| Claude Sonnet 5.5claude-sonnet-5-5 | 63.9 % | — | Google DeepMind Gemini 4 Hydrogen Model Card evaluation table (2026-11-12) | Vendor-reported |
| Agents' Last Examv1.1 (Google Nov 2026 card) | ||||
| Gemini 4 Hydrogengemini-4-hydrogen | 71.6 % | — | Google DeepMind Gemini 4 Hydrogen Model Card evaluation table (2026-11-12) | Vendor-reported |
| Claude Sonnet 5.5claude-sonnet-5-5 | 60.4 % | — | Google DeepMind Gemini 4 Hydrogen Model Card evaluation table (2026-11-12) | Vendor-reported |
| Vending-Bench 2 (Bank balance)Google Nov 2026 card | ||||
| Gemini 4 Hydrogengemini-4-hydrogen | 14549 USD | — | Google DeepMind Gemini 4 Hydrogen Model Card evaluation table (2026-11-12) | Vendor-reported |
| Claude Sonnet 5.5claude-sonnet-5-5 | 11165 USD | — | Google DeepMind Gemini 4 Hydrogen Model Card evaluation table (2026-11-12) | Vendor-reported |
| OSWorld (2.1 (offline subset, Google Nov 2026 card))2.1 (offline subset, Google Nov 2026 card) | ||||
| Gemini 4 Hydrogengemini-4-hydrogen | 77.6 % | — | Google DeepMind Gemini 4 Hydrogen Model Card evaluation table (2026-11-12) | Vendor-reported |
| Claude Sonnet 5.5claude-sonnet-5-5 | 83.9 % | — | Google DeepMind Gemini 4 Hydrogen Model Card evaluation table (2026-11-12) | Vendor-reported |
| Terminal-Bench-Science (0.1 (Google Nov 2026 card))0.1 (Google Nov 2026 card) | ||||
| Gemini 4 Hydrogengemini-4-hydrogen | 51.4 % | — | Google DeepMind Gemini 4 Hydrogen Model Card evaluation table (2026-11-12) | Vendor-reported |
| Claude Sonnet 5.5claude-sonnet-5-5 | 59.9 % | — | Google DeepMind Gemini 4 Hydrogen Model Card evaluation table (2026-11-12) | Vendor-reported |
| DataAnalyticsBench | ||||
| Gemini 4 Hydrogengemini-4-hydrogen | 43 questions correct | — | Plotly Studio Embedded through OpenRouter, one run of 43 private analytics questions. | Third-party |
| DataAnalyticsBench full-run inference cost | ||||
| Gemini 4 Hydrogengemini-4-hydrogen | 6.36 USD per 43-question run | — | Plotly reported inference cost at OpenRouter prices on the run date. | Third-party |
Each row records who published and who ran it (see the Harness column); numbers compare only within a documented run.
Sources: Anthropic — Introducing Claude Haiku 5.5 (2026-10-07) · Artificial Analysis (independent) — Independent same-harness comparison: Gemini 4 Hydrogen vs Claude Sonnet 5.5 (Intelligence Index v5.0.1) · Google DeepMind — Gemini 4 Hydrogen Model Card (November 2026 evaluation table vs Claude Sonnet 5.5) · Plotly — Gemini 4 Hydrogen Benchmark Results for Data Analysis
06 // BY WORKLOAD
Which model for which workload
Confidence labels: Measured = same-harness evidence · Price-derived = list-price arithmetic · Inferred = reasoned from docs · Insufficient evidence = no comparable public data.
| Workload | Our pick | Confidence | Why |
|---|---|---|---|
| Budget-sensitive / high volume | Gemini 4 Hydrogen | Measured | 25% cheaper tokens, $0 cache writes vs $2.50, 3.2x cheaper AA top-tier cost ($3.16 vs $10.02), and $0.80/task at medium effort (Index 41 vs Sonnet medium $0.78 Index 31). |
| Agent workflows | Gemini 4 Hydrogen | Measured | Hydrogen leads AutomationBench-AA (76.1% vs 71.8%), Vending-Bench 2 ($14,549 vs $11,165), and Agents' Last Exam v1.1 (71.6% vs 60.4%), making it stronger on SaaS/API tool loops; pick Sonnet for GUI OSWorld (83.9% vs 77.6%). |
| Complex repository maintenance | Claude Sonnet 5.5 | Measured | Despite Hydrogen winning SWE-sweep (13.5% vs 3.1%), Sonnet 5.5 keeps a +10.1pp lead on Terminal-Bench 4.0 (63.6% vs 53.5% AA; 63.9% vs 57.0% Google TB 4.1) and +2.7pp on FrontierCode-AA 2.0 (41.7% vs 39.0%). |
| Daily coding tasks | Gemini 4 Hydrogen | Measured | At everyday High effort, Hydrogen scores Index 51 ($1.87/task) vs Sonnet High Index 42 ($1.64/task), while costing zero on cache writes across frequent repo context refreshes. |
| Long-context tasks | Gemini 4 Hydrogen | Price-derived | Both keep flat 1M input pricing, but Hydrogen charges $0 for implicit cache writes (saving $2.50/MTok vs Sonnet on shifting prefixes) and supports 1,048,576-token background outputs vs Sonnet's 128K cap. |
| Latency-sensitive production | Gemini 4 Hydrogen | Measured | Hydrogen streams faster (172 vs 147 tok/s) and finishes AA top-effort tasks in 14.9 min vs Sonnet Max's 32.3 min, with faster TTFA (42.88s vs 295.79s). |
| Quality-first | Decide by bake-off | Measured | Near-tie overall (Sonnet Max 55 vs Hydrogen Xhigh 54), sharply split by domain: Sonnet for AA-Briefcase office docs (1815 vs 1568) & Terminal-Bench; Hydrogen for Omniscience (46 vs 30), HLE-Diamond (58.8% vs 53.3%), and CritPt v2 (66.0% vs 59.0%). |
| Offline batch jobs | Gemini 4 Hydrogen | Price-derived | Both offer 50% Batch ($0.75/$3.75 vs $1/$5), and Hydrogen also offers synchronous Flex at the same $0.75/$3.75 price without waiting on async batch queues. |
HOLD CRITERIA
When staying on the lower-cost or incumbent model is reasonable
- Your agents heavily rely on Terminal-Bench-style bash/shell coding recovery (63.6% vs 53.5% AA) or desktop GUI automation (OSWorld 83.9% vs 77.6%).
- You produce structured financial/legal deliverables matching AA-Briefcase (1815 vs 1568 Elo).
- Your prompt prefixes turn over rapidly with strict cache-control TTL requirements served better by explicit Anthropic cache writes.
EVALUATION PROTOCOL
4-step Gemini 4 Hydrogen vs Claude Sonnet 5.5 trial
Public benchmarks don't run your tasks. Before committing a fleet, measure both models on your own work.
- 1Set calculator inputs with cache-write tokens = 0 on Hydrogen (implicit) vs real write churn on Sonnet ($2.50/MTok) to reflect true cache-miss costs.
- 2Run your top 25 production prompts through Hydrogen at 'high' ($1.87 AA avg) and 'xhigh' ($3.16 AA avg) against your current Sonnet effort tier.
- 3Monitor bash/CLI retry loops separately — Sonnet 5.5 retains a ~10pp Terminal-Bench advantage even though Hydrogen wins AutomationBench-AA and Omniscience.
- 4Test Gemini Flex ($0.75/$3.75) for latency-tolerant queues where you previously had to rewrite pipelines for Batch API.
QUESTIONS
Frequently asked questions
Why does Cache Write cost $0 on Gemini 4 Hydrogen vs $2.50/MTok on Sonnet 5.5?
Gemini 4 Hydrogen uses Google's implicit prompt caching automatically (95% discount on cache hits = $0.075/MTok) with no explicit cache-creation surcharge. Sonnet 5.5 requires explicit cache checkpoints billed at 1.25x ($2.50/MTok for 5m TTL, 2x for 1h TTL).
Why is Gemini 4 Hydrogen compared at 'xhigh' against Sonnet 5.5 at 'max' on AA?
Gemini 4 Hydrogen exposes thinking levels low, medium, high, and xhigh (no 'max' level exists on Hydrogen), whereas Claude Sonnet 5.5 exposes up to 'max'. Even one nominal effort tier lower, Hydrogen Xhigh scores 54 vs Sonnet Max 55 while costing $3.16 vs $10.02/task.
When is Claude Sonnet 5.5 still clearly better than Gemini 4 Hydrogen?
In both AA v5.0.1 and DeepMind's own model card, Sonnet 5.5 maintains clear leads on CLI/shell recovery (Terminal-Bench 63.6% vs 53.5% AA; 63.9% vs 57.0% TB 4.1), GUI computer use (OSWorld 83.9% vs 77.6%), and office deliverables (AA-Briefcase 1815 vs 1568 Elo).
CONTINUE EXPLORING
Related
VERIFICATION LOG
Sources & freshness
- Googleaccessed 2026-11-19Gemini 4 Hydrogen API documentation (gemini-4-hydrogen specs, 1M window, thinking levels)https://aistudio.google.com/docs/models/gemini-4-hydrogen · published 2026-11-12Input, thinking and output share a 1,048,576-token window. Synchronous output defaults to 65,536 tokens. Hydrogen supports background long decode and implicit caching.
- Anthropicaccessed 2026-10-09Claude models overview (model IDs, context window, max output, retirement)https://platform.claude.com/docs/en/docs/about-claude/models/overviewclaude-opus-5-5: 1M context, 128K max output, training cutoff Jun 2026, retirement not sooner than 2027-09-22, cache reads '5% of input price', batch 50% off.
- Googleaccessed 2026-11-18Introducing Gemini 4 Hydrogen for Agentic Workflows (GA announcement)https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-hydrogen/ · published 2026-11-12Nov 12, 2026 Google announcement introducing Gemini 4 Hydrogen ($1.50/$7.50 per MTok flat across 1M context, Free tier available) targeting high intelligence per dollar in coding and agentic tool workflows.
- Anthropicaccessed 2026-11-19Claude pricing (API pricing table incl. cache write/read, fast mode, batch)https://claude.com/pricingOpus 5.5: $4/$20 per MTok, cache read $0.20, cache write (5m TTL) $5, batch 'save 50%', fast mode 2x on input/output. Re-verified 2026-11-18: Claude Fable 5.2 ($8/$40, $0.20 cached read = 2.5%, $10.00 5m write).
- Anthropicaccessed 2026-11-19Claude pricing docs (cache TTL prices, fast/batch stacking rules)https://platform.claude.com/docs/en/about-claude/pricingOpus 5.5 1h cache write $8/MTok; caching multipliers apply on top of fast-mode pricing and stack with the Batch discount; fast mode not available with Batch. Re-verified 2026-11-18: Fable 5.2 Batch 50% off ($4/$20), Fast 2x ($16/$80), 2.5% cache-read stacking, 1h cache write $16.00.
- Googleaccessed 2026-11-19Gemini Developer API pricinghttps://aistudio.google.com/docs/pricingThe paid tables distinguish fresh input, implicit cache hits and output including thinking. Argon and Carbon have introductory Standard, Batch and Flex rates through November 30, 2026, with the next prices effective December 1. Google lists Priority separately.
- Anthropicaccessed 2026-10-09Introducing Claude Haiku 5.5 (2026-10-07)https://www.anthropic.com/claude-haiku-5-5 · published 2026-10-07Tiered pricing ($0.10/$0.50 up to 100K, $0.50/$2.50 over) with Haiku 4.5 comparison ('90% lower up to 100K', 'on average ~75% less to run'); benchmark table includes GPT-6 Luna rows (GDPval-AA v2.1, AA-Briefcase v1.1, OSWorld 2.1, Terminal-Bench 4.0, FrontierCode 1.1, Chartography); 'fastest model to date' (at standard speed, slower than Opus in Fast Mode); tokenizer caveat 'uses slightly more tokens per task'.
- Artificial Analysis (independent)accessed 2026-11-18Independent same-harness comparison: Gemini 4 Hydrogen vs Claude Sonnet 5.5 (Intelligence Index v5.0.1)https://artificialanalysis.ai/models/comparisons/gemini-4-hydrogen-vs-claude-sonnet-5-5 · published 2026-11-18Independent same-harness Intelligence Index v5.0.1 pair comparison: Gemini 4 Hydrogen (xhigh: Index 54, $3.16/task, 14.9 min, 172 tok/s; high: Index 51, $1.87/task) vs Claude Sonnet 5.5 (max: Index 55, $10.02/task, 32.3 min, 147 tok/s; xhigh: Index 50, $4.09/task).
- Google DeepMindaccessed 2026-11-19Gemini 4 Hydrogen Model Card (November 2026 evaluation table vs Claude Sonnet 5.5)https://deepmind.google/models/model-cards/gemini-4-hydrogen/ · published 2026-11-12DeepMind identifies Hydrogen as adapted from Carbon. The November 12 evaluation table compares Hydrogen with Claude Sonnet 5.5 and GPT-6.1 Terra.
- Plotlyaccessed 2026-11-19Gemini 4 Hydrogen Benchmark Results for Data Analysishttps://plotly.com/blog/gemini-4-hydrogen-plotly-data-analytics-bench/ · published 2026-11-12Plotly reported Hydrogen's score and cost for the same private analytics exam on November 12.
Pricing data last verified 2026-11-19. Vendors ship frequently — re-verify before making spend commitments. See the methodology page for update cadence and stale-data flags.