BenchLM comparison
ChatGPT vs Gemini (July 2026): Benchmarks, Price & Verdict
Gemini wins on benchmarked scores and price: Gemini 3.1 Proposts 89 on BenchLM versus 80 for GPT-5.5, ChatGPT's last fully scored flagship, at less than half the API price. ChatGPT counters with the brand-new GPT-5.6 family, being benchmarked now with standout early agentic results (91.9 on Terminal-Bench 2).
This is a comparison of ChatGPT and Gemini as products, so we anchor it to the flagship models behind each app: ChatGPT runs on OpenAI's GPT-5.6 family — Sol, Terra, and Luna(GA on July 9, 2026), while Gemini runs on Google's Gemini 3.1 Pro and Gemini 3.5 Flash. Every number below is BenchLM's sourced benchmark or pricing data for those models.
Search it as "ChatGPT vs Gemini" or "Gemini vs ChatGPT" — either way, this is the highest-volume rivalry in consumer AI, and in July 2026 the data gives a more interesting answer than the market-share numbers suggest.
The short answer
On fully benchmarked models, Gemini is ahead. Gemini 3.1 Pro scores 89 overall on BenchLM — nine points clear of GPT-5.5's 80 — and it also leads on Arena Elo, 1485.8 to 1474.62. Even Google's fast tier holds up: Gemini 3.5 Flash scores 75 with a 1479.5 Arena Elo, remarkable for a budget model.
The complication is that ChatGPT just changed its engine. The GPT-5.6 family shipped on July 9, 2026, and none of its three models have an overall BenchLM score yet. The early sourced results for GPT-5.6 Solare strong precisely where Gemini 3.1 Pro is weakest: 91.9 on Terminal-Bench 2 versus Gemini's 77, and 92.2 on BrowseComp versus 86. If agentic work is your use case, the verdict may flip once Sol is fully scored.
On price, Google is the aggressor: Gemini 3.1 Pro costs $2 input / $12 output per million tokens against $5 / $30 for GPT-5.6 Sol. Today, Gemini is the better-scoring, cheaper choice; ChatGPT is the fresher stack with the stronger agentic profile and the more mature consumer app.
ChatGPT vs Gemini: the models behind the apps
Here are the flagship models side by side. BenchLM scores are our sourced overall benchmark aggregate; Arena Elo is the human-preference rating where published. The GPT-5.6 models read "benchmarking now" because sourced coverage is still filling in after their July 9 launch.
| Model | BenchLM score | Arena Elo | Context |
|---|---|---|---|
| GPT-5.6 SolOpenAI | Benchmarking now | — | 1M |
| GPT-5.6 TerraOpenAI | Benchmarking now | — | 1M |
| GPT-5.6 LunaOpenAI | Benchmarking now | — | 1M |
| GPT-5.5OpenAI | 80 | 1474.6 | 1M |
| Gemini 3.1 ProGoogle | 89 | 1485.8 | 1M |
| Gemini 3.5 FlashGoogle | 75 | 1479.5 | 1M |
For the benchmark-by-benchmark detail, see the dedicated model pages: Gemini 3.1 Pro vs GPT-5.6 Sol for the current flagship matchup, or Gemini 3.1 Pro vs GPT-5.5 for the fully scored comparison.
ChatGPT vs Gemini pricing: subscriptions and API
ChatGPT Plus costs $20/monthand is the reference consumer AI subscription. Google sells Gemini access through its paid Google AI subscription tiers, which bundle the Gemini app's top models with Workspace and storage perks — a different shape of deal that is hard to compare line-for-line, and Google also gives away more capability at the free tier than OpenAI does.
The API numbers are directly comparable. Google wins decisively at the flagship tier — Gemini 3.1 Pro costs less than half of GPT-5.6 Sol per token — while at the budget tier OpenAI's GPT-5.6 Luna ($1 / $6) actually undercuts Gemini 3.5 Flash ($1.50 / $9). Full rate cards live on the OpenAI API pricing and Google API pricing hubs.
| Model | Input / 1M tokens | Output / 1M tokens | Note |
|---|---|---|---|
| GPT-5.6 Sol | $5.00 | $30.00 | — |
| GPT-5.6 Terra | $2.50 | $15.00 | — |
| GPT-5.6 Luna | $1.00 | $6.00 | — |
| GPT-5.5 | $5.00 | $30.00 | Previous ChatGPT flagship, still on the API |
| Gemini 3.1 Pro | $2.00 | $12.00 | Rises to $4 / $18 above 200K-token prompts |
| Gemini 3.5 Flash | $1.50 | $9.00 | — |
Which should you pick?
Pick ChatGPT if…
- Agentic and terminal work is the job: GPT-5.6 Sol's early 91.9 on Terminal-Bench 2 crushes Gemini 3.1 Pro's 77.
- You lean on web-research agents — Sol's 92.2 BrowseComp leads Gemini 3.1 Pro's 86.
- You want the newest flagship family: GPT-5.6 shipped July 9, 2026, a full generation fresher than Gemini 3.1 Pro.
- The ChatGPT ecosystem matters — custom GPTs, voice, and the most mature plugin and memory surface in consumer AI.
- You want a predictable $20/month subscription rather than a bundle tied to Google's ecosystem.
Pick Gemini if…
- You want the highest fully benchmarked model in this matchup: Gemini 3.1 Pro's 89 overall leads GPT-5.5's 80 by nine points.
- Price matters at the top end: $2 / $12 per million tokens for 3.1 Pro versus $5 / $30 for GPT-5.6 Sol — a fraction of the cost for the higher benchmarked score.
- Human preference is your proxy for quality — Gemini 3.1 Pro's 1485.8 Arena Elo tops every scored OpenAI model on this page.
- You live in Google's ecosystem: Gemini is wired into Gmail, Docs, and Android in ways ChatGPT cannot match.
- You want strong math and reasoning on a budget — 3.1 Pro posts 97 on MATH-500 and 92 on BBH.
How this comparison works
BenchLM compares products through the models that power them — "ChatGPT" and "Gemini" each swap in new flagship models several times a year, so a product verdict is only as current as the models behind it. Overall scores are our sourced benchmark aggregate; Arena Elo is the community human-preference rating where one has been published; API prices come from each provider's official rate card as recorded in BenchLM's pricing data. Where a model is too new to have an overall score — the entire GPT-5.6 family right now — we say so instead of guessing, and cite only the category results that have landed. Every individual benchmark quoted in the text comes from the same sourced dataset that powers our model comparison pages.
ChatGPT vs Gemini FAQ
Is Gemini better than ChatGPT?
On today's fully benchmarked models, yes: Gemini 3.1 Pro scores 89 on BenchLM against 80 for GPT-5.5, and leads on Arena Elo (1485.8 vs 1474.6). The caveat is that ChatGPT's GPT-5.6 family is brand new and still being benchmarked — its early agentic results are the strongest in this matchup, so treat the verdict as provisional until Sol's overall score lands.
Is Gemini cheaper than ChatGPT?
At the flagship tier, yes: Gemini 3.1 Pro costs $2 input / $12 output per million tokens — less than half of GPT-5.6 Sol's $5 / $30 — despite the higher benchmarked score. At the budget tier the order flips: GPT-5.6 Luna ($1 / $6) is cheaper than Gemini 3.5 Flash ($1.50 / $9). On subscriptions, ChatGPT Plus is a flat $20/month while Google bundles Gemini into its paid AI tiers, with a generous free tier below.
Which is better for coding, ChatGPT or Gemini?
ChatGPT, on the current sourced data. GPT-5.6 Sol's early coding results — 91.9 on Terminal-Bench 2 and a 77.4 Artificial Analysis coding index — lead Gemini 3.1 Pro's 77 and 68.8 on the same measures, and even the older GPT-5.5 beats it on Terminal-Bench 2 (82 vs 77). Gemini 3.1 Pro fights back on SWE-bench Verified (75) and shines on math-heavy work.
Can I use both ChatGPT and Gemini?
Absolutely — and Gemini's free tier makes this the cheapest "both" combo of the big three. A sensible split: keep ChatGPT Plus for agentic tasks, browsing research, and its ecosystem, and use Gemini free or paid for long-document work, math, and anything living in Gmail or Docs. On the API, routing bulk workloads to Gemini 3.5 Flash while reserving Sol for hard agentic jobs is a common cost play.