Josh McGrath

Post-Training Researcher, OpenAI · 1 appearance on the record.

computed by AI from the episodes · how this works → · full disclaimer →

Josh McGrath is an AI researcher at OpenAI who has contributed to the post-training of models including GPT-4o, o1, and o3. His work focuses on reinforcement learning with verifiable rewards, post-training data curation, token efficiency, and agent capabilities.

11statements → 5claims → 3claims resolved → 33%fully supported → 3.36/5average certainty → 2/5average debate potential →

1 supported 2 partly supported 0 contradicted 2 not checkable as stated how the 5 claims stand · each chip opens the sources

2 predictions · 3 assertions · 5 insights · 1 disclosure · every statement was checked. The predictions and assertions are the 5 claims: statements the public record can support or contradict. 3 are resolved, and 2 name no date, number or outcome precise enough to check. Everything else (opinions, insights, what ifs, disclosures) can never be settled by the record, so it carries no assessment.

The record, in short

What the tape says about how Josh argues and how the claims held up. Everything they said, and everything said about them, is in the tabs below.

Their most notable supported claim

Assertion Supported

McGrath: GPT-5 Thinking Matches or Beats Deep Research on Published Evals

“I mean, I think if you look at our published evals, they're, they look, like, basically on par if it's not better, so, like, I mean, that's personally what I do.”

Josh McGrath Dec 31, 2025 ▶ 6:46 [State of Post-Training] From GPT-4.1 to 5.1: RLVR, Agent & Token Efficiency — Josh McGrath, OpenAI

Expressed certainty vs assessment result

none yet certainty 1

none yet certainty 2

75% certainty 3

50% certainty 4

none yet certainty 5

weighted support: a fully supported claim counts one, a partly supported claim counts half. Each filled bar is clickable and opens exactly those claims; "none yet" means nothing said at that certainty level has resolved yet

What they said11 Episodes1

Everything Josh McGrath said on Latent Space that made the record, most notable first. Filter by type, assessment or year in the ledger →

Prediction Not checkable as stated

McGrath: Specialized and Frontier Reasoning AI Models Will Eventually Converge

“You know, I think if you look at like deep research, the original one and GPT-Five thinking on like high reasoning today, I think you'll see that like eventually the models all sort of converge in their capabilities.”

Josh McGrath Dec 31, 2025 ▶ 6:13 [State of Post-Training] From GPT-4.1 to 5.1: RLVR, Agent & Token Efficiency — Josh McGrath, OpenAI

Insight

McGrath: RLHF and RLVR differ by data quality, not optimization math

“Really, at the end of the day, like, RLHF, RLVR, They're both policy gradient methods, but the, what's different is just like the input data.”

Josh McGrath Dec 31, 2025 ▶ 9:02 [State of Post-Training] From GPT-4.1 to 5.1: RLVR, Agent & Token Efficiency — Josh McGrath, OpenAI

Insight

McGrath: DeepSeek Math's real breakthrough is verifiable reward trust, not GRPO

“As you said, it came out in the deep seek math paper, and like, it's an interesting optimization method, but it's like the more interesting thing that they have a new reward signal that they sort of like re that we can really, really trust. Like when, you know…”

Josh McGrath Dec 31, 2025 ▶ 12:42 [State of Post-Training] From GPT-4.1 to 5.1: RLVR, Agent & Token Efficiency — Josh McGrath, OpenAI

Insight

McGrath: RL runs have far more infrastructure failure points than pre-training

“The issue with RL is, like, you're doing tasks, and each task could have, like, a different grading setup, and each one of those different grading setups, that's, like, more infrastructure, and so, You know, when I'm staying up late trying to figure out what's…”

Josh McGrath Dec 31, 2025 ▶ 2:12 [State of Post-Training] From GPT-4.1 to 5.1: RLVR, Agent & Token Efficiency — Josh McGrath, OpenAI

Insight

McGrath: Design specs let Codex complete hours of coding in 15 minutes

“If I spend, like, you know, 30, 40 minutes writing something that looks like a design doc or something, Codex can do more work than I can do in a few hours in, like, 15 minutes.”

Josh McGrath Dec 31, 2025 ▶ 3:50 [State of Post-Training] From GPT-4.1 to 5.1: RLVR, Agent & Token Efficiency — Josh McGrath, OpenAI

Assertion Supported

McGrath: GPT-5 Thinking Matches or Beats Deep Research on Published Evals

“I mean, I think if you look at our published evals, they're, they look, like, basically on par if it's not better, so, like, I mean, that's personally what I do.”

Josh McGrath Dec 31, 2025 ▶ 6:46 [State of Post-Training] From GPT-4.1 to 5.1: RLVR, Agent & Token Efficiency — Josh McGrath, OpenAI

Assertion Partly supported

McGrath: GPT-5.1 dramatically reduced token usage over GPT-5 while boosting evals

“Yeah, and so you can see, like, from five to 5.1, our overall evals, you know, we bumped some. But if you look at a two D plot of how many tokens it takes for us to get that, it went way down.”

Josh McGrath Dec 31, 2025 ▶ 13:58 [State of Post-Training] From GPT-4.1 to 5.1: RLVR, Agent & Token Efficiency — Josh McGrath, OpenAI

Prediction Not checkable as stated

McGrath: AGI will be a single tool that decides its own thinking time

“Yeah, I think, like, eventually, you know, we'll have AGI, and like, you're not gonna have to worry too much about how hard to think directly. It'll just, you know, we'll have a one tool that you always go to, and it knows how long to think for, and things lik…”

Josh McGrath Dec 31, 2025 ▶ 15:23 [State of Post-Training] From GPT-4.1 to 5.1: RLVR, Agent & Token Efficiency — Josh McGrath, OpenAI

Insight

McGrath: AI Frontier Is Bottlenecked by Shortage of Hybrid Systems-ML Talent

“I think we're still having trouble not at OpenAI, but I think as a whole, producing lots of people that do lot, want to do lots of both systems work and ML work. And I think if you're trying to push the frontier, you don't know which Place is currently bottlen…”

Josh McGrath Dec 31, 2025 ▶ 21:36 [State of Post-Training] From GPT-4.1 to 5.1: RLVR, Agent & Token Efficiency — Josh McGrath, OpenAI

Assertion Partly supported

McGrath: OpenAI 10xed Effective Context Window for GPT-4.1

“I worked on long context, that was why I was on last, was for 4.1, where we, you know, I think, tenxed the effective context window for 4.1”

Josh McGrath Dec 31, 2025 ▶ 16:40 [State of Post-Training] From GPT-4.1 to 5.1: RLVR, Agent & Token Efficiency — Josh McGrath, OpenAI

Disclosure

McGrath: OpenAI Continues to Release Non-Thinking Models for Specific APIs

“No, we're still, we still are releasing non-thinking models but that one was the one that we did that was like API-specific non-thinking so, you know, focus has shifted a little.”

Josh McGrath Dec 31, 2025 ▶ 0:47 [State of Post-Training] From GPT-4.1 to 5.1: RLVR, Agent & Token Efficiency — Josh McGrath, OpenAI

Appearances (1)

Episode	Date	Speaking time
[State of Post-Training] From GPT-4.1 to 5.1: RLVR, Agent & Token Efficiency — Josh McGrat	Dec 31, 2025	15m

Josh McGrath

The record, in short

Their most notable supported claim

Expressed certainty vs assessment result

Appearances (1)

Turn any episode into a week of clips.