Google originally planned to launch Gemini 3.5 Pro this month.
The Gemini Pro models do not seem to be iterating anywhere near as quickly as Claude or GPT (last release was 3.1 Pro in February). Its causing a growing performance gap between Google and the other two labs, and the Gemini 3.5 Flash model, good as it is, doesn't close it much.
Many users criticize Google's complacency and slow iteration on Gemini as a culture problem causing it to lag behind Claude and GPT, while some express hope that the upcoming 3.5 Pro will outperform competitors.
As you can see
It does seem like meaningfully better AI releases are accelerating, especially from OpenAI & Anthropic.
To illustrate, I caused this timeline to be created. It only lists new models that scored 3 points or higher over previous models in the Artificial Analysis index.
@AndrewCurran_ The velocity point remains either way.
@emollick They said at I/O that Gemini 3.5 Pro would arrive this month, so in the next three weeks. I think the reason for the delay is that it is under safety review.
@emollick They said at I/O that Gemini 3.5 Pro would arrive this month, so in the next three weeks. I think the reason for the delay is that it is under safety review.
The Gemini Pro models do not seem to be iterating anywhere near as quickly as Claude or GPT (last release was 3.1 Pro in February). Its causing a growing performance gap between Google and the other two labs, and the Gemini 3.5 Flash model, good as it is, doesn't close it much.
Nobody paid attention to how strong Gemini Flash 3.5 is in Antigravity. I always hated the Gemini models for coding, but now I choose it easily over Opus 4.6 in Antigravity. I also don’t believe that 4.8 is significantly better than 4.6. Anthropic regressed in a few important things. Flash 3.5 is the best model I have worked with for agentic tasks. It just runs and is extremely fast. The token flat fees are very generous. Curious to see if the next Pro model will be as good. For now I use it for execution, Kimi 2.6 for planning, and DeepSeek v4 Pro for preparing the documentation of large codebases. Everyone who still pays the absurd prices of Misanthropic has lost control of his life. Also: Droid is the better CLI harness compared to CC.
@emollick Iteration speed is a culture problem, not a compute problem. Claude and GPT ship like startups. Gemini ships like Google.
@emollick @AndrewCurran_ Any insights from Google on why? Feels almost intentional?
@emollick This just isnt true at all. The anthropic models actually being served are way worse than 3.1 pro and 3.5 flash.
@mc_code_ @emollick Or another way to see it, making calculated bets on where to utilize the capacity.
@emollick Google's distribution moat is turning into a complacency trap
@emollick The model gap shows up fastest in the small daily annoyances. Not leaderboard scores, but whether the model can hold context across a messy task without making you re-explain the obvious. Once users feel that friction for a week, “good enough” stops being good enough.
@emollick They don't want frontier models - just good enough to work with their consumer products. Risky bet.
@emollick Only thing Gemini is good at is getting metadata from youtube videos and of course notebookLM
@AndrewCurran_ @emollick sorry its just wasn't that good i think
@emollick Do you think Google will be able to catch up? Or will the gap increase the closer we get to RSI?
@emollick The release cadence gap matters a lot for founders deciding which stack to build on. When you're integrating at the application layer, betting on a model that's not keeping pace creates compounding tech debt. The switching cost is non-trivial.
The Pro lag is more interesting than the Flash comparison. Flash iterations are cheap; Pro model training is expensive. Google might be deliberately slowing the Pro frontier cycle to divert compute toward Flash inference at scale. The distribution moat (Android, Workspace) needs reliability and cost, not frontier bragging rights.
@emollick Been saying so long Google needs a massive layoff to dump the org inefficiencies
@emollick Curious if this is a resourcing choice or a strategy. Google has the talent — maybe they're saving the real jump for one big release instead of steady iteration. Türkçe: Bu bir kaynak tercihi mi yoksa strateji mi merak ediyorum. 😅
@emollick In our experience the open source Gemma 4 has been quite good though. Maybe the gap is in the harnesses. I don't user Gemini Pro tbh. Codex and to lesser extent Claude Code are quite good.
@emollick i keep forgetting 3.1 pro is still the current gemini pro, feels like it dropped longer than four months ago
Google originally planned to launch Gemini 3.5 Pro this month.
The Gemini Pro models do not seem to be iterating anywhere near as quickly as Claude or GPT (last release was 3.1 Pro in February). Its causing a growing performance gap between Google and the other two labs, and the Gemini 3.5 Flash model, good as it is, doesn't close it much.