Gemini 4 Neon
Launch tracker · day 2
Gemini 4 Neon at launch: the first independent results
As of 19 Nov 2026, Gemini 4 Neon (Google, released 18 Nov 2026) has independent results from Artificial Analysis (AA Intelligence Index v5.0.1 59.6), LMArena (Arena Text 1519), LiveBench (LiveBench 2.0 87.4%) and Scale AI (Humanity's Sixth Sense 75.9%). It ranks #5 of 343 models on the BenchLeader Index, at 67.3. It is listed at $3.00 in / $15.00 out per 1M tokens, with a 1M-token context window. Not yet measured by Epoch AI, Vals AI, ARC Prize and SWE-bench.
- Released
- 18 Nov 2026 (day 2)
- Provider
- Price
- $3.00 in / $15.00 out per 1M tokens
- Context window
- 1M tokens
- Open weights
- No
- BenchLeader Index
- 67.3 (#5 of 343 models)
| Independent board | Results so far |
|---|---|
| Artificial Analysis | AA Intelligence Index v5.0.1 59.6 · AA-Omniscience v1.1 53.8 · HLE-Diamond (AA) 68.4% · more below |
| LMArena | Arena Text 1519 · Arena Hard Prompts 1546 · Arena Coding 1558 · more below |
| LiveBench | LiveBench 2.0 87.4% · LiveBench 2.0 Reasoning 89.6% · LiveBench 2.0 Coding 89.7% · more below |
| Scale AI | Humanity's Sixth Sense 75.9% |
Boards usually post results within days of a launch; this page is rebuilt every three hours.
Launch comparisons
Scores from independent publishers only; benchmark figures a provider publishes about its own model are not shown.
Better value: GPT-6.1 Sol scores 65.9 for $4.00/M, 33% cheaper.
No cheaper ranked model scores higher: it is on the price frontier.
As of 19 Nov 2026, Gemini 4 Neon by Google scores 67.3 on the BenchLeader Index (top 10 of 631 ranked configurations), with its best results in agents & tools (76) and weakest in human preference (72). API pricing: $3.00 per 1M input tokens and $15.00 per 1M output tokens ($6.00 blended at 3:1), list price on 19 Nov 2026. It streams about 100 tokens/s (Artificial Analysis). Released 18 Nov 2026.
Key facts (as of 19 Nov 2026)
- Gemini 4 Neon API pricing: $3.00 per 1M input tokens and $15.00 per 1M output tokens ($6.00 blended at 3:1) (list price, 19 Nov 2026).
- Gemini 4 Neon streams about 100 tokens/s (Artificial Analysis, 19 Nov 2026).
- Gemini 4 Neon was released on 18 Nov 2026 (Google).
In detail. Gemini 4 Neon is a Google proprietary reasoning model, released 18 Nov 2026. Its best configuration (high reasoning effort) ranks #7 of 631 on the BenchLeader Index at 67.3 ±2.8, in the top ten. The ± is the point: 44 other configurations score within that range, so they and this one cannot be told apart on quality alone — price and speed are what separate them. It scores highest in agents & tools (76) and lowest in human preference (72). At $6.00 per million tokens blended it is among the most expensive ranked models. Output speed of 101 tokens per second puts it faster than most, with a first answer in 40.6 s. It has been measured at 4 reasoning-effort settings; this summary describes the best-scoring one, and the tabs above switch between them. Last measured 18 Nov 2026.
- Blended price
- $6.00/M
- $3.00 in · $15.00 out
- via Artificial Analysis
- Output speed
- 101 tok/s
- measured by Artificial Analysis
- First answer
- 41 s
- first token 2.66 s
- Context
- 1.0M
- via models.dev
- Accepts
- Text · Images · Audio · Video · PDFs
- via models.dev, OpenRouter
- Cost per task
- $3.74
- 15 min a task, AA Index
- Tokens per task
- 90k
- 51% of it reasoning
- Released
- 18 Nov 2026
- Overall index67
- Reasoning72
- Coding73
- Agents & tools76
- Knowledge72
- Human preference72
Embed this rank
More badge styles[](https://www.benchleader.com/models/gemini-4-neon)Paste it at the top of README.md, beside your other badges. The badge redraws with every data update and links back to this page.
Reasoning-effort configurations
The same model behaves differently depending on how much it is allowed to think. Each row is one setting, scored only on the benchmarks that were run at that setting. “Not stated” collects results from publishers that did not say which setting they used; for a reasoning model that is usually its thinking mode, but we do not assume it. Pick a setting here or at the top of the page to see its price, speed and category scores.
| Effort | Index | Rank | Speed | First answer | Chat reply cost | Categories |
|---|---|---|---|---|---|---|
| low | – | – | 97 tok/s | 14 s | $0.0057 | Coding 56 · Knowledge 70 · Reasoning 68 |
| medium | – | – | 101 tok/s | 20 s | $0.0057 | Coding 66 · Knowledge 71 · Reasoning 71 |
| highbest | 67.3 | #7 | 101 tok/s | 41 s | $0.0057 | Agents & tools 76 · Coding 73 · Human preference 72 · Knowledge 72 · Reasoning 72 |
| xhigh | 64.3 | #30 | 110 tok/s | 57 s | $0.0057 | Coding 73 · Composite 72 · Knowledge 73 · Reasoning 74 |
Benchmark results
See all LLM benchmarks. One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.
Reasoning
| Benchmark | low | medium | high | xhigh | Source | Trend |
|---|---|---|---|---|---|---|
| Arena Hard Prompts | – | – | 1546#1 | – | Arena (LMArena) | |
| LiveBench 2.0 Reasoningnot in index | – | – | – | 89.6%#17 | LiveBench | |
| HLE-Diamond (AA) | 54.2%#32 | 63.6%#16 | 68.4%#7 | 71.3%#5 | Artificial Analysis | |
| CritPt v2 hard subset (AA) | 62.0%#29 | 61.5%#31 | 59.0%#44 | 60.5%#38 | Artificial Analysis | |
| Arena Expertnot in index | – | – | 1541#6 | – | Arena (LMArena) |
Coding
| Benchmark | low | medium | high | xhigh | Source | Trend |
|---|---|---|---|---|---|---|
| Arena Coding | – | – | 1558#2 | – | Arena (LMArena) | |
| Arena WebDev | – | – | 1764#6 | – | Arena (LMArena) | |
| LiveBench 2.0 Codingnot in index | – | – | – | 89.7%#3 | LiveBench | |
| Terminal-Bench 4.0 (AA)not in index | 32.8%#74 | 54.5%#32 | 61.6%#8 | 63.6%#5 | Artificial Analysis | |
| FrontierCode 2.0 (AA) | 26.9%#60 | 36.9%#30 | 43.3%#12 | 44.4%#9 | Artificial Analysis |
Agents & tools
| Benchmark | low | medium | high | xhigh | Source | Trend |
|---|---|---|---|---|---|---|
| Arena Agent | – | – | 18.4#3 | – | Arena (LMArena) | |
| LiveBench Long Agenticnot in index | – | – | – | 82.6%#2 | LiveBench | |
| LiveBench Browsernot in index | – | – | – | 95.6%#1 | LiveBench | |
| Arena Agent Task Completionnot in index | – | – | 19.2#2 | – | Arena (LMArena) | |
| AutomationBenchnot in index | 76.0%#12 | 77.8%#4 | 79.2%#2 | 79.3%#1 | Artificial Analysis | |
| GDP.pdf (AA)not in index | 31.4%#8 | 32.8%#2 | 33.4%#1 | 28.0%#34 | Artificial Analysis | |
| AA-Briefcase v1.1not in index | 1297#108 | 1518#55 | 1632#20 | 1699#8 | Artificial Analysis | |
| Terminal-Bench-Science 0.1 (AA)not in index | 27.1%#53 | 54.8%#24 | 65.7%#12 | 70.0%#6 | Artificial Analysis | |
| MKA (AA)not in index | 43.4%#15 | 50.9%#9 | 54.6%#3 | 56.5%#2 | Artificial Analysis |
Knowledge
| Benchmark | low | medium | high | xhigh | Source | Trend |
|---|---|---|---|---|---|---|
| LiveBench 2.0 Data Analysisnot in index | – | – | – | 85.3%#2 | LiveBench | |
| LiveBench Researchnot in index | – | – | – | 94.5%#1 | LiveBench | |
| AA-Omniscience v1.1 | 48.5#11 | 51.1#9 | 53.8#7 | 54.8#6 | Artificial Analysis | |
| AA-Omniscience v1.1: accuracynot in index | 56.6%#40 | 57.9%#39 | 59.4%#33 | 60.5%#29 | Artificial Analysis | |
| AA-Omniscience v1.1: non-hallucinationnot in index | 81.3%#18 | 83.8%#12 | 86.1%#9 | 85.6%#10 | Artificial Analysis |
Instruction following
| Benchmark | high | Source | Trend |
|---|---|---|---|
| Arena Instruction Followingnot in index | 1522#3 | Arena (LMArena) |
Human preference
| Benchmark | high | Source | Trend |
|---|---|---|---|
| Arena Text | 1519#3 | Arena (LMArena) | |
| Arena Creative Writingnot in index | 1514#4 | Arena (LMArena) | |
| Arena Multi-turnnot in index | 1529#3 | Arena (LMArena) | |
| Arena Longer Queriesnot in index | 1521#6 | Arena (LMArena) |
Multimodal
| Benchmark | xhigh | Source | Trend |
|---|---|---|---|
| LiveBench Designnot in index | 81.5%#5 | LiveBench | |
| Humanity's Sixth Sensenot in index | 75.9%#1 | Scale AI SEAL |
Long context
| Benchmark | low | medium | high | xhigh | Source | Trend |
|---|---|---|---|---|---|---|
| MLCRnot in index | 26.1%#43 | 30.6%#28 | 28.9%#34 | 31.7%#24 | Artificial Analysis |
Composite
| Benchmark | low | medium | high | xhigh | Source | Trend |
|---|---|---|---|---|---|---|
| LiveBench 2.0 | – | – | – | 87.4%#1 | LiveBench | |
| AA Intelligence Index v5.0.1not in index | 44.1#48 | 54.9#20 | 59.6#9 | 61.4#6 | Artificial Analysis |
Where it wins
Benchmarks where this configuration ranks in the top five of every configuration measured.
API pricing: where to run it
Every provider serving this model through OpenRouter, with throughput and first-token latency measured on live traffic over the last 30 minutes and each provider’s own price. Purple marks the best in each column.
| Provider | Speed | First token | Input $/M | Output $/M | Context | Quantisation |
|---|---|---|---|---|---|---|
| Google Vertex | 117 tok/s | 23 s | $5.40 | $27.00 | 1.0M | – |
| Google Vertex | 88 tok/s | 3.35 s | $3.00 | $15.00 | 1.0M | – |
| Google AI Studio | 85 tok/s | 2.42 s | $3.00 | $15.00 | 1.0M | – |
| Google AI Studio | 70 tok/s | 5.03 s | $1.50 | $7.50 | 1.0M | – |
| Google Vertex | 63 tok/s | 2.94 s | $3.30 | $16.50 | 1.0M | – |
What a task costs
Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache at $0.150 per 1M. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0057 | $0.0048 | 43.6 s |
| Summarize a 30-page report | 12,000 / 600 | $0.045 | $0.019 | 46.6 s |
| Code edit | 6,000 / 1,500 | $0.041 | $0.028 | 55.4 s |
| Agentic coding session | 60,000 / 4,000 | $0.240 | $0.112 | 1.3 min |
| Structured extraction | 2,000 / 200 | $0.0090 | $0.0047 | 42.6 s |
More from Google
Compare Gemini 4 Neon with…
- Gemini 4 Neon vs GPT-6 Astra
- Gemini 4 Neon vs Claude Opus 5.5
- Gemini 4 Neon vs Gemini 4 Carbon
- Gemini 4 Neon vs Claude Fable 5.1
- Gemini 4 Neon vs GPT-6.1 Sol
- Gemini 4 Neon vs GPT-6.2 Astra
- Gemini 4 Neon vs Claude Fable 5
- Gemini 4 Neon vs GPT-5.6 Sol
- Gemini 4 Neon vs Claude Opus 5
- Gemini 4 Neon vs Gemini 4 Argon
- Gemini 4 Neon vs GPT-6.2 Sol
- Gemini 4 Neon vs Muse Wonder
- Gemini 4 Neon vs Muse Spark
- Gemini 4 Neon vs GPT-5.5
- Gemini 4 Neon vs Gemini 4 Hydrogen
- Gemini 4 Neon vs Claude Sonnet 5.5
- Gemini 4 Neon vs GPT-6.1 Terra
- Gemini 4 Neon vs Claude Fable 5.2
- Gemini 4 Neon vs Claude Opus 4.7
- Gemini 4 Neon vs GPT-5.4
- Gemini 4 Neon vs Claude Opus 4.6
- Gemini 4 Neon vs Gemini 3.5 Flash
- Gemini 4 Neon vs GPT-6 Sol
- Gemini 4 Neon vs Qwen3.6 Max
- Gemini 4 Neon vs Gemini 3.7 Flash
- Gemini 4 Neon vs Muse Spark 1.1
- Gemini 4 Neon vs Muse Spark 1.2
- Gemini 4 Neon vs Muse Wonder Full
- Gemini 4 Neon vs Kimi K3.1 Pro
- Gemini 4 Neon vs Gemma 4 12B
See also
Embed or cite
Show where Gemini 4 Neon ranks on a model card or README. The badge redraws itself with every daily update, carries the data date and links back to this page.
[](https://www.benchleader.com/models/gemini-4-neon)Cite this
The ranking changes daily, so a reference carries the data date. APA style:
BenchLeader. (2026, November 19). Gemini 4 Neon on the BenchLeader Index (method v1.5) [Data set]. Retrieved November 19, 2026, from https://www.benchleader.com/models/gemini-4-neon
BibTeX:
@misc{benchleader_gemini_4_neon_20261119,
author = {{BenchLeader}},
title = {{Gemini 4 Neon on the BenchLeader Index (method v1.5)}},
year = {2026},
month = nov,
howpublished = {\url{https://www.benchleader.com/models/gemini-4-neon}},
note = {Data as of 2026-11-19}
}Every badge style, the embeddable top-10 widget and how to cite the Index
Questions people ask
- What are Gemini 4 Neon's benchmark scores?
- As of 19 Nov 2026, Gemini 4 Neon (high effort) scores 67.3 ±2.8 on the BenchLeader Index, top 10, strongest in agents & tools (76) and weakest in human preference (72). That rests on independent benchmark results only, and every score on this page links to its publisher.
- How does Gemini 4 Neon rank?
- Gemini 4 Neon (high effort) is #7 of 631 ranked configurations on 19 Nov 2026 (top 10). 44 other configurations score within its margin, so the exact position matters less than the band. The leader is Claude Fable 5.2 (max effort) at 71.5.
- How much does Gemini 4 Neon cost per 1M tokens?
- As of 19 Nov 2026, Gemini 4 Neon costs $3.00 per 1M input tokens and $15.00 per 1M output tokens, $6.00 blended (three input tokens per output token).
- Is Gemini 4 Neon better than GPT-6 Astra?
- As of 19 Nov 2026, Gemini 4 Neon (high) and GPT-6 Astra (max) are neck and neck on quality (67.3 vs 66.9 on the BenchLeader Index, inside the ±3.3 margin of error), and Gemini 4 Neon (high) costs 3.3× less, which makes it the better buy for most workloads. The head-to-head page has the benchmark-by-benchmark detail.
- When was Gemini 4 Neon released?
- Google released Gemini 4 Neon on 18 Nov 2026. BenchLeader has results for it at 4 reasoning-effort settings.
Data as of 19 Nov 2026. Compare these configurations.
Cite as: BenchLeader, “Gemini 4 Neon: benchmarks, pricing, speed and rank”, data as of 19 Nov 2026, https://www.benchleader.com/models/gemini-4-neon.