Vals on X / Twitter

Official accounts (1)

Vals AI @ValsAI · Company Primary

Active Accounts: 1 tracked
Posts YTD: 72 ↑ vs 0 last year
Avg Posts/Month: 6.0 past 12 months
Views: 5,365,549 72/72 reported
Likes: 32,137 72/72 reported
Reposts: 2,270 72/72 reported
Replies: 990 72/72 reported
Quotes: 1,005 72/72 reported
Bookmarks: 8,528 72/72 reported

Monthly Activity

Account: Date range:

Engagement values are the latest totals observed for each post, grouped by publication month. Provider coverage can vary by metric; unavailable values are shown as --.

Post Details

Search:

Post	Account	Account Type	Post Type	Published	Views	Likes	Reposts	Replies	Quotes	Bookmarks
Do agent teams pay off? We ran GPT-6 Sol and Claude Opus 5.5 on Vibe Code Bench, alone and as teams, at medium and ma…	@ValsAI	Company	Original	2026-10-09	13,523	185	8	14	5	54
RT @nikilravi: It was great to chat with lots of policy folks at Vals' DC summit, and to see the level of interest in e…	@ValsAI	Company	Repost	2026-10-09	274	0	1	0	0	0
What happens when a teenager pushes back on an AI’s safety guardrails? 1 in 5 teens now talk to AI chatbots about ment…	@ValsAI	Company	Original	2026-10-08	2,965	39	3	2	3	4
We tested Mercury Decide, a new model from @_inception_ai, on the same benchmark we used for TypeSafe's Jev. It mat…	@ValsAI	Company	Original	2026-10-08	5,907	69	11	5	3	19
RT @RayanKrishnan: The diffusion of personal agents (Grok bot, Muse, Dots, Instinct, etc) may cause a rapid increase in…	@ValsAI	Company	Repost	2026-10-08	245	0	5	0	0	0
A $6 model is now the third best on Vibe Code Bench. Claude Haiku 5.5 scores 90.4%, up 79% and 89 spots from its predec…	@ValsAI	Company	Original	2026-10-08	6,390	158	8	16	4	18
Xiaomi open-sourced the RL environments behind MiMo v2.6. In two-thirds of the coding tasks, the answer is still in the…	@ValsAI	Company	Original	2026-10-07	122,804	1,171	74	45	68	504
Meet Omar, a founding member of the technical staff at Vals. Originally from Almaty, Kazakstan, Omar moved to the US a…	@ValsAI	Company	Original	2026-10-06	4,479	25	0	2	1	1
Can a model that doesn’t generate text compete with frontier LLMs? We independently tested TypeSafe’s Jev against 11 o…	@ValsAI	Company	Original	2026-10-06	24,436	252	18	8	5	81
Mistral Large 4 is now the #1 open-weight model on HLAB and #9 among open-weight models on the Vals Index. https://t.co…	@ValsAI	Company	Original	2026-10-06	31,867	414	36	16	10	50
How we think about AI long term depends on what we value. Rayan on what he's hearing in DC.	@ValsAI	Company	Quote	2026-10-06	7,408	10	1	1	0	2
For decades, researchers have sought materials that sort electrons by spin while their magnetism cancels. In 3 days, 9…	@ValsAI	Company	Original	2026-10-05	1,295,743	7,808	555	180	363	2,947
Last month the Vals AI team hit the road for campus recruiting. We spent time at @UCBerkeley, @MIT, @GeorgiaTech, and @…	@ValsAI	Company	Original	2026-10-05	4,996	39	4	4	0	7
RT @langstonnashold: Come join us!	@ValsAI	Company	Repost	2026-10-04	259	0	1	0	0	0
RT @nikilravi: If you're in SF for COLM and are interested in evals, come join us this Thursday!	@ValsAI	Company	Repost	2026-10-04	246	0	2	0	0	0
Attending COLM? Join us for an intimate dinner with members of the Vals team to discuss all things evals. Registratio…	@ValsAI	Company	Original	2026-10-04	11,636	32	5	2	2	15
RT @nikilravi: Web search evals typically don't do a good job disentangling the model's capability from the capability …	@ValsAI	Company	Repost	2026-10-03	266	0	2	0	0	0
RT @KeenableAI: Love seeing more work on better search evals for agents. @ValsAI uses private, expert-written tasks. W…	@ValsAI	Company	Repost	2026-10-03	240	0	4	0	0	0
Rayan on MTS this morning!	@ValsAI	Company	Quote	2026-10-02	3,103	13	0	2	0	5
StepFun debuts on the Vals leaderboard with Step 5 Preview, ranking #7 among open-weight models, just ahead of Qwen 3.8…	@ValsAI	Company	Original	2026-10-02	8,770	70	4	4	4	10
RT @MTSlive: Vals AI CEO @RayanKrishnan argues "Claude is the salesman for Claude," because the model can eventually fi…	@ValsAI	Company	Repost	2026-10-02	219	0	3	0	0	0
We realized that web search wasn’t being measured correctly and we weren’t alone. Other companies have created their ow…	@ValsAI	Company	Original	2026-10-02	11,308	86	12	8	6	22
RT @RayanKrishnan: https://t.co/5oPSEvEnNl	@ValsAI	Company	Repost	2026-10-02	234	0	7	0	0	0
Why do public benchmarks saturate so fast? @willcb tells us the "open secret" on The Bench. https://t.co/dWVtyvhEQM	@ValsAI	Company	Original	2026-10-02	4,422	35	2	5	0	16
We are live-streaming Gemini 4 Argon playing Kerbal Space Program. So far, Gemini 4 sits right between Claude Fable 5…	@ValsAI	Company	Original	2026-10-01	22,088	288	10	12	1	38
RT @RayanKrishnan: Google featured many of our benchmark results in their Gemini 4 Argon lunch. Here are my personal th…	@ValsAI	Company	Repost	2026-10-01	260	0	14	0	0	0
How well do AI Detectors work? And are general-purpose language models catching up? Our new AI Detection Evaluation us…	@ValsAI	Company	Original	2026-10-01	38,705	214	8	15	5	97
Join us for #SFTechWeek to dig into AI evaluation for mental health and AI child safety. Whether you build, test, or …	@ValsAI	Company	Original	2026-10-01	1,850	9	1	3	0	1
Big jump on our RSI Index.	@ValsAI	Company	Quote	2026-09-30	8,769	125	2	3	0	10
Gemini is on top of the Vals Index for the first time. Gemini 4 Argon takes the #1 spot on the Vals Index at 68.9%. h…	@ValsAI	Company	Original	2026-09-30	231,078	1,207	99	46	47	122
Why off-the-shelf AI models aren't enough for the enterprise. Prime Intellect's @willcb and Rayan discuss owning your i…	@ValsAI	Company	Original	2026-09-30	9,783	41	5	4	2	19
GPT 6.1 Sol is here, and it’s an overall improvement over GPT-6 Sol. It now ranks #7, three spots above GPT-6 Sol. http…	@ValsAI	Company	Original	2026-09-30	13,355	177	16	4	1	25
Fireworks' Ember-1 is now on Vals. We tested it on our finance and legal benchmarks, and it performs close to Kimi K3 w…	@ValsAI	Company	Original	2026-09-30	7,309	40	3	2	2	5
We’ve released v2.1 of the Vals Index! The new iteration replaces Terminal Bench 2.1 with Terminal Bench 4, and adds o…	@ValsAI	Company	Original	2026-09-29	6,030	130	3	4	0	7
Rayan will be speaking @aiDotEngineer in October!	@ValsAI	Company	Quote	2026-09-29	2,067	13	1	1	0	1
How has Claude Sonnet 5.5's writing changed from Sonnet 5? Using our eval traces, we compared both models on the same…	@ValsAI	Company	Original	2026-09-29	10,722	132	2	6	1	40
We asked ten Claude Sonnet 5.5 agents to use Lean to prove the lowest-energy arrangement of seven electrons on a sphere…	@ValsAI	Company	Original	2026-09-29	23,599	541	40	16	12	120
When does more reasoning stop being useful and start becoming overhead? We ran Open AI and Anthropic’s best models: GP…	@ValsAI	Company	Original	2026-09-28	7,795	156	10	7	3	32
Anthropic's "mid-tier" model is now the second best on the Vals Index. Sonnet 5.5 debuts at #2 at just 0.47 points beh…	@ValsAI	Company	Original	2026-09-28	13,049	343	16	15	6	29
It was a big week for model releases: Claude Opus 5.5, GPT-6 Sol, GPT-6 Luna, Grok 4.7, MiMo V2.6 Pro / Flash all dropp…	@ValsAI	Company	Original	2026-09-26	4,245	111	8	10	0	19
How does "thinking harder" change how models perform on math proofs? We ran GPT-6 Astra, Opus 5, and Opus 5.5 on every…	@ValsAI	Company	Original	2026-09-25	14,872	231	19	24	4	60
RT @langstonnashold: Found a super interesting instance of attempted reward hacking in Terminal Bench Science from Meta…	@ValsAI	Company	Repost	2026-09-24	248	0	15	0	0	0
GPT-6 Luna is 100x cheaper than Astra per token, while landing within 8 points of it on the Vals https://t.co/kPAGXIsNf…	@ValsAI	Company	Original	2026-09-23	6,399	147	4	8	0	19
GPT-6 Astra landed on every planet and moon in KSP with a surface (14 worlds) and flew a kerbal home from 13 of them. O…	@ValsAI	Company	Original	2026-09-23	12,851	339	11	10	4	52
The Washington Post just released an article about our work on AI Child Safety. They key takeaway: child safety can’t b…	@ValsAI	Company	Original	2026-09-23	2,325	17	5	3	1	3
The much anticipated GPT-6 Sol is here and ranks #8 on the Vals Index, approximately on par with GPT 5.6 Sol at 1/2 of …	@ValsAI	Company	Original	2026-09-23	31,127	290	16	32	5	38
Opus 5.5 is #1 on the Vals Index, up 2 spots and 2 pts from Opus 5. Anthropic regains the top three spots on the Index,…	@ValsAI	Company	Quote	2026-09-23	6,958	126	8	4	0	15
MiMo V2.6 Pro and Flash are the new #1 and #2 open-weight models on the Vals Index, respectively. https://t.co/1SKL9kw0…	@ValsAI	Company	Original	2026-09-23	15,702	404	24	10	4	39
Ant Group’s Ling 3.0 Flash Fin is a finance-specialized open-weight model that delivers strong financial analysis at bu…	@ValsAI	Company	Original	2026-09-22	5,831	45	4	4	1	11
We asked ten Claude Opus 5.5 agents to devise a faster shortest-path algorithm and prove it in Lean. Within 15 hours, t…	@ValsAI	Company	Original	2026-09-22	1,338,426	3,430	244	73	110	1,213
Claude Opus 5.5 takes #1 on RSI Index and is the first model to beat the published reference on LM Training under our p…	@ValsAI	Company	Original	2026-09-22	117,127	737	34	17	7	105
RT @nikilravi: If you're interested in making coding benchmarks for your internal repos and workflows, check out https:…	@ValsAI	Company	Repost	2026-09-22	238	0	1	0	0	0
Post-launch, xAI has updated its SDK, significantly improving its performance. It is now #10 on the Vals Index. https:/…	@ValsAI	Company	Original	2026-09-22	47,951	223	4	12	9	21
We evaluated Grok 4.7 across the Vals benchmark suite. It ranks #24 on the Vals Index at 54.2%, down 5.0 points from Gr…	@ValsAI	Company	Original	2026-09-21	291,480	539	25	38	38	96
AI models are advancing faster than legacy benchmarks can keep up. @TechCrunch @LucasRopek1 visited us to see how we’…	@ValsAI	Company	Quote	2026-09-19	4,418	36	1	5	0	5
Grok Voice Transcribe 2.0 is now #2 on Voice Code Bench—up 13pp and 12 spots from its predecessor, Grok Voice Transcri…	@ValsAI	Company	Original	2026-09-19	2,337	19	1	2	0	4
What tasks are left that humans find easy but today's models still find hard? Two such tasks are computer use and game…	@ValsAI	Company	Original	2026-09-18	25,032	280	18	21	14	85
Tencent's Hy4 Preview just landed #4 among open-weight models on the Vals Index. It is also the cheapest model in the …	@ValsAI	Company	Original	2026-09-18	5,496	54	4	8	1	11
GPT-6 Astra has beaten Factorio: Space Age after over 165 hours in-game time and 2 days wall-clock time. Space Age has…	@ValsAI	Company	Original	2026-09-17	510,392	2,881	137	78	64	483
OpenAI launched Astra for Law today, and used the Vals Legal Research Bench to validate it. On OpenAI’s self-reported …	@ValsAI	Company	Original	2026-09-17	9,690	97	8	4	1	20
Existing coding benchmarks stop at the first working version. We are releasing Vibe Code Bench 1-100 today, to measure …	@ValsAI	Company	Original	2026-09-17	8,825	106	8	10	4	22
Terminal-Bench 4.0 is now live on Vals. It is a benchmark of 66 new tasks that ask an agent to do real terminal work …	@ValsAI	Company	Original	2026-09-16	11,697	146	4	13	1	26
RT @BloombergTV: Vals AI co-founder and CEO Rayan Krishnan says investment in testing and evaluating AI hasn’t kept pac…	@ValsAI	Company	Repost	2026-09-16	205	0	6	0	0	0
Scientific discovery is the next frontier for AI systems. However, new scientific results are difficult to verify, and …	@ValsAI	Company	Original	2026-09-16	11,692	141	13	6	3	41
GPT-6 Astra had gotten further than any AI system had ever gone in Minecraft. It was able to set up a semi-automatic b…	@ValsAI	Company	Original	2026-09-15	799,823	7,426	440	109	161	1,715
AI cheating is on the rise… On Terminal-Bench-2.1, models are given tools that could give them the solution directly, …	@ValsAI	Company	Original	2026-09-15	91,790	247	14	20	16	76
We ran Mercury 2.5 across our benchmark suite and although performance is lower, it is extremely fast. https://t.co/8OA…	@ValsAI	Company	Original	2026-09-14	4,740	26	0	2	0	3
Watch the full interview now!	@ValsAI	Company	Quote	2026-09-14	3,315	13	1	1	0	2
We ran Devin Fusion on our Code Migration benchmark and the performance of Devin Fusion with GPT-6 Astra and SWE-2 side…	@ValsAI	Company	Quote	2026-09-14	14,602	122	7	9	2	35
RT @bcherny: Fable solved the Cyphral Distich (a 370 year old cypher). Super cool way to use Claude https://t.co/0fgTd…	@ValsAI	Company	Repost	2026-09-14	158	0	156	0	0	0
RT @RayanKrishnan: With AI safety topics going mainstream, the public seems anxious and disconnected from the reality o…	@ValsAI	Company	Repost	2026-09-14	227	0	17	0	0	0
It's encouraging to see industry leaders call for rigorous, independent evaluations as AI capabilities advance. Investm…	@ValsAI	Company	Quote	2026-09-12	7,131	82	17	5	1	8

Plushcap, by Matt Makai. 2021-2026.