2026-11-18 — Gemini 4 Neon Resets Coding Cost Frontier
Viewing 2026-11-18 snapshot Latest →
01
Quick Picks
Pick a workflow for default, budget, and max options
Default NEW default
Google
CursorBench 4.1
63.7%
Per task (Cursor)
$6.54
$3/$15 Output 62k
Hallucination 13.9%
Why Ranks 2nd on CursorBench (63.7%) at 55% of Opus 5.5 task spend.
Caveat Trails Opus by 0.6 pts on pure terminal tasks; peak SWE relies on official CLI.
Budget NEW lowest bill
GPT-6.2 Sol
maxOpenAI
Terminal-Bench 4.1
61.5%
Per task (AA)
$1.01
$2/$10 Output 40k
Hallucination 55.4%
Why 40k output budget produces the lowest per-task spend for routine scripts.
Caveat Input above 272K triggers higher pricing; 55.4% open-fact hallucination rate.
Max NEW #1 (65.4%)
Gemini 4 Neon
xhighGoogle
CursorBench 4.1
65.4%
Per task (Cursor)
$10.67
$3/$15 Output 142k
Hallucination 14.4%
Why Highest multi-file refactor score with 65.2% on Terminal-Bench.
Caveat Costs 63% more per task than high tier for a 1.7 pt score gain.
Agent Synthesis Key Changes
01 Gemini 4 Neon takes the top two spots on CursorBench (65.4% / 63.7%). Its high tier averages $6.54 per task, 45% below Opus 5.5. 02 GPT-6.2 Astra and Sol compress per-task output to ~40k tokens. Despite an $8/$40 list price, Astra averages $4.00 per AA task. 03 Gemini 4 Carbon introductory half-price API ($2/$10) ends Nov 30, reverting to $4/$20 on Dec 1. Read (3) Hide
Key Changes
01 Gemini 4 Neon takes the top two spots on CursorBench (65.4% / 63.7%). Its high tier averages $6.54 per task, 45% below Opus 5.5. 02 GPT-6.2 Astra and Sol compress per-task output to ~40k tokens. Despite an $8/$40 list price, Astra averages $4.00 per AA task. 03 Gemini 4 Carbon introductory half-price API ($2/$10) ends Nov 30, reverting to $4/$20 on Dec 1.
Initial baseline
- 01
Gemini 4 Neon takes the top two spots on CursorBench (65.4% / 63.7%). Its high tier averages $6.54 per task, 45% below Opus 5.5.
- 02
GPT-6.2 Astra and Sol compress per-task output to ~40k tokens. Despite an $8/$40 list price, Astra averages $4.00 per AA task.
- 03
Gemini 4 Carbon introductory half-price API ($2/$10) ends Nov 30, reverting to $4/$20 on Dec 1.
02
Cost & Capability
Hidden thinking tokens drive the real bill, not list prices
COST RANGE 26×
$1.01 →
$26.54
Sol max 57.0 Fable 5.2 max 65.0
RATE VS REAL BILL 3.1× cheaper
Sol max $2/$10
$1.01 40k
Hydrogen xhigh $1.5/$7.5
$3.16 110k
List ×1.3 · Bill $1.01 vs $3.16
CODING LEAD 65.4%
Neon xhigh 65.4% · $10.67
Neon high 63.7% · $6.54
Fable 5.2 max · AA 65.0 · $26.54
Pareto frontier: Sol max ($1.01) → Neon high ($3.74) → Astra max ($4.00) → Fable 5.2 max ($26.54).
Hover or tap nodeThinking Tokens & Bills
Claude Sonnet 5.5 max
206k + 79k = 285k Claude Fable 5.2 max
193k + 81k = 274k Claude Opus 5.5 max
120k + 53k = 173k Gemini 4 Neon xhigh
81k + 61k = 142k Gemini 4 Neon high
35k + 27k = 62k GPT-6.2 Astra max
26k + 15k = 41k GPT-6.2 Sol max
23k + 17k = 40k AA v5.0.1 averages (thinking + answer); bills include input and cache.
03
Full Benchmarks
Raw scores and per-task spend, filterable and sortable
Direct Sources · 9 models · Click column to sort
| | |||||||
|---|---|---|---|---|---|---|---|
| Claude Fable 5.2
◈
max | $8 / $40 | 65.0 | $26.54 | 61.3%
$48.11 | 65.8%
| — | 59.3% |
| GPT-6.2 Astra
◈
max | $8 / $40 ≤272K | 62.0 | $4.00 | — | 62.1%
| — 6.0: 44.0% | 59.1% |
| Gemini 4 Neon xhigh | $3 / $15 | 61.0 | $6.44 | 65.4%
$10.67 | 65.2%
| — | 14.4% |
| Gemini 4 Neon
◈
high | $3 / $15 | 60.0 | $3.74 | 63.7%
$6.54 | — xhigh 65.2% | 46.2% | 13.9% |
| Gemini 4 Carbon high | $2 / $10* Dec 1: $4/$20 | 60.0 | $4.90 | 61.5%
$8.52 | 64.5%
| — | 11.4% |
| Claude Opus 5.5 max | $4 / $20 | 59.0 | $10.04 | 58.1%
$11.88 | 65.8%
| 47.3% | 50.3% |
| GPT-6.2 Sol
◈
max | $2 / $10 ≤272K | 57.0 | $1.01 | — | 61.5%
| 43.8% | 55.4% |
| Claude Sonnet 5.5 max | $2 / $10 | 55.0 | $10.02 | 54.4%
$8.63 | 63.9%
| — | — |
| Gemini 4 Hydrogen xhigh | $1.5 / $7.5 | 54.0 | $3.16 | 61.4%
$8.09 | 56.4%
high | — | 8.5% |
| No open-weights models were recorded in this early snapshot. | |||||||