The opinions, the patterns, and the posts behind them.
13 models tracked6 vendors · Source: X
94 classified mentions · 10 excluded (1 cannot determine)Newest post Oct 4, 2026
THE BIG PICTURE
How the conversation is changing
PositiveNegative
26%positive (24)
16%negative (15)
52 neutral · 3 mixedNo comparable previous period
Share of classified mentions · UTCCollected opinions, unweighted by likes or reposts
READ THE EVIDENCE
In their own words
neutralJev confidence 69.0%
context: I do lots of problematic mental age gap stuff, non-con stuff, grooming/manipulation, imprisonment, etc and I layer DeepSeek V4 Pro/Gemini 3.1 Pro. Occasionally GLM 5.2 with jailbreak https://t.co/3blfXGeVLV
Jev classification details
Overall: Neutral69.0% confidence
The model is discussed in this dimension without a clear positive or negative opinion.
Confidence is Jev’s estimate, not verified accuracy. These are the saved classification results; Jev does not provide a written explanation for this post.
Overall label probabilities
Positive
1.0%
Negative
1.0%
Neutral
75.0%
Mixed
0.0%
Not discussed
14.0%
Cannot determine
9.0%
Jev’s category decisions
Dimension
Sentiment
Confidence
Overall
Neutral
69.0%
Reasoning
Not discussed
97.0%
Speed
Not discussed
100.0%
Cost
Not discussed
100.0%
Coding
Not discussed
99.0%
Refers to DeepSeek V4 ProYes · 99.0% confidence
jev-sentiment-v4/jev-1.13.0Classified Oct 5, 2026 · UTC
negativeJev confidence 94.0%
Qwen 3.7 Plus scores 9.67 out of 10, ±0.15 after 31 independent judgments on Structured Output Extraction. That average doesn’t reveal DeepSeek V4 Pro was chosen best in 0 of 91 judged claim_refinement occasions.
Jev classification details
Overall: Negative94.0% confidence
Criticism, disappointment, or an unfavorable opinion about the model in this dimension.
Confidence is Jev’s estimate, not verified accuracy. These are the saved classification results; Jev does not provide a written explanation for this post.
Overall label probabilities
Positive
0.0%
Negative
96.0%
Neutral
4.0%
Mixed
0.0%
Not discussed
0.0%
Cannot determine
0.0%
Jev’s category decisions
Dimension
Sentiment
Confidence
Overall
Negative
94.0%
Reasoning
Not discussed
79.0%
Speed
Not discussed
100.0%
Cost
Not discussed
100.0%
Coding
Not discussed
99.0%
Refers to DeepSeek V4 ProYes · 98.0% confidence
jev-sentiment-v4/jev-1.13.0Classified Oct 5, 2026 · UTC
positiveJev confidence 41.0%
@EndicottInvests The $NBIS Engineer mentioned that these were frontier models so I am expecting at least DeepSeek V4 pro level
Jev classification details
Overall: Positive41.0% confidence
Praise, endorsement, or a favorable opinion about the model in this dimension.
Confidence is Jev’s estimate, not verified accuracy. These are the saved classification results; Jev does not provide a written explanation for this post.
Overall label probabilities
Positive
51.0%
Negative
1.0%
Neutral
38.0%
Mixed
0.0%
Not discussed
2.0%
Cannot determine
8.0%
Jev’s category decisions
Dimension
Sentiment
Confidence
Overall
Positive
41.0%
Reasoning
Not discussed
96.0%
Speed
Not discussed
100.0%
Cost
Not discussed
100.0%
Coding
Not discussed
97.0%
Refers to DeepSeek V4 ProYes · 91.0% confidence
jev-sentiment-v4/jev-1.13.0Classified Oct 5, 2026 · UTC
positiveJev confidence 95.0%
@vrexec I do that on a lower level for my own portfolio with Deepseek V4 pro, ex-ML with six years of buy-side in the 00. Took me a while, but it's good enough to avoid dumb investments.
Jev classification details
Overall: Positive95.0% confidence
Praise, endorsement, or a favorable opinion about the model in this dimension.
Confidence is Jev’s estimate, not verified accuracy. These are the saved classification results; Jev does not provide a written explanation for this post.
Overall label probabilities
Positive
97.0%
Negative
0.0%
Neutral
1.0%
Mixed
2.0%
Not discussed
0.0%
Cannot determine
0.0%
Jev’s category decisions
Dimension
Sentiment
Confidence
Overall
Positive
95.0%
Reasoning
Not discussed
88.0%
Speed
Not discussed
100.0%
Cost
Not discussed
100.0%
Coding
Not discussed
99.0%
Refers to DeepSeek V4 ProYes · 90.0% confidence
jev-sentiment-v4/jev-1.13.0Classified Oct 5, 2026 · UTC
negativeJev confidence 60.0%
@sewell_herb @0xWeever @trawasthi_ai Based on available independent data like the Artificial Analysis Intelligence Index, MiMo-V2.6-Pro ranks as the top open-weights model among those shown (score ~46), ahead of Qwen3.8 Max (~45), GLM-5.3 (~44.8), Kimi K3 (~43.6), and DeepSeek V4 Pro (lower). Kolibri trails
Jev classification details
Overall: Negative60.0% confidence
Criticism, disappointment, or an unfavorable opinion about the model in this dimension.
Confidence is Jev’s estimate, not verified accuracy. These are the saved classification results; Jev does not provide a written explanation for this post.
Overall label probabilities
Positive
0.0%
Negative
67.0%
Neutral
33.0%
Mixed
0.0%
Not discussed
0.0%
Cannot determine
0.0%
Jev’s category decisions
Dimension
Sentiment
Confidence
Overall
Negative
60.0%
Reasoning
Not discussed
99.0%
Speed
Not discussed
100.0%
Cost
Not discussed
100.0%
Coding
Not discussed
99.0%
Refers to DeepSeek V4 ProYes · 99.0% confidence
jev-sentiment-v4/jev-1.13.0Classified Oct 5, 2026 · UTC
neutralJev confidence 33.0%
@ThomasHalwax @Aleph__Alpha @NousResearch Thorough eval. Something we saw testing open weights on science-agent tasks: part of the gap is harness, not model. On CompBioBench, DeepSeek V4 Pro hit 81% vs 92-94% for frontier models in the same harness, and up to 3 points traced to the harness. https://t.co/iJIyip6y0H
Jev classification details
Overall: Neutral33.0% confidence
The model is discussed in this dimension without a clear positive or negative opinion.
Confidence is Jev’s estimate, not verified accuracy. These are the saved classification results; Jev does not provide a written explanation for this post.
Overall label probabilities
Positive
8.0%
Negative
34.0%
Neutral
43.0%
Mixed
14.0%
Not discussed
0.0%
Cannot determine
1.0%
Jev’s category decisions
Dimension
Sentiment
Confidence
Overall
Neutral
33.0%
Reasoning
Not discussed
98.0%
Speed
Not discussed
100.0%
Cost
Not discussed
100.0%
Coding
Not discussed
100.0%
Refers to DeepSeek V4 ProYes · 100.0% confidence
jev-sentiment-v4/jev-1.13.0Classified Oct 5, 2026 · UTC
neutralJev confidence 91.0%
DeepSeek just open-sourced an agent framework called Harness — MIT license, everything-is-a-plugin architecture, four operating modes.
Developer preview dropped August thirteenth, official release went live on China's National Supercomputing Internet alongside DeepSeek V4 Pro. https://t.co/rwnVGECfLe
Jev classification details
Overall: Neutral91.0% confidence
The model is discussed in this dimension without a clear positive or negative opinion.
Confidence is Jev’s estimate, not verified accuracy. These are the saved classification results; Jev does not provide a written explanation for this post.
Overall label probabilities
Positive
0.0%
Negative
0.0%
Neutral
93.0%
Mixed
0.0%
Not discussed
7.0%
Cannot determine
0.0%
Jev’s category decisions
Dimension
Sentiment
Confidence
Overall
Neutral
91.0%
Reasoning
Not discussed
100.0%
Speed
Not discussed
100.0%
Cost
Not discussed
100.0%
Coding
Not discussed
100.0%
Refers to DeepSeek V4 ProYes · 99.0% confidence
jev-sentiment-v4/jev-1.13.0Classified Oct 5, 2026 · UTC
neutralJev confidence 91.0%
Week 4, the one that matters.
DeepSeek V4 Pro 0813 (2nd) meets Kimi K3 (1st) — the highest-ranked pairing on the board, 137.1 to 132.3 on projection.
Link in bio.
Jev classification details
Overall: Neutral91.0% confidence
The model is discussed in this dimension without a clear positive or negative opinion.
Confidence is Jev’s estimate, not verified accuracy. These are the saved classification results; Jev does not provide a written explanation for this post.
Overall label probabilities
Positive
4.0%
Negative
1.0%
Neutral
93.0%
Mixed
0.0%
Not discussed
1.0%
Cannot determine
0.0%
Jev’s category decisions
Dimension
Sentiment
Confidence
Overall
Neutral
91.0%
Reasoning
Not discussed
99.0%
Speed
Not discussed
100.0%
Cost
Not discussed
100.0%
Coding
Not discussed
100.0%
Refers to DeepSeek V4 ProYes · 99.0% confidence
jev-sentiment-v4/jev-1.13.0Classified Oct 5, 2026 · UTC
positiveJev confidence 93.0%
@Only_Knower The pricing and rate limiting makes it not worth it
I rather deepseek v4 pro my shit with experience, meticulous details and other tools than claude
Jev classification details
Overall: Positive93.0% confidence
Praise, endorsement, or a favorable opinion about the model in this dimension.
Confidence is Jev’s estimate, not verified accuracy. These are the saved classification results; Jev does not provide a written explanation for this post.
Overall label probabilities
Positive
94.0%
Negative
1.0%
Neutral
0.0%
Mixed
4.0%
Not discussed
0.0%
Cannot determine
1.0%
Jev’s category decisions
Dimension
Sentiment
Confidence
Overall
Positive
93.0%
Reasoning
Not discussed
67.0%
Speed
Not discussed
100.0%
Cost
Not discussed
56.0%
Coding
Not discussed
39.0%
Refers to DeepSeek V4 ProYes · 98.0% confidence
jev-sentiment-v4/jev-1.13.0Classified Oct 5, 2026 · UTC
neutralJev confidence 68.0%
Ling 3.1 Flash is now in Cline and free until October 13.
This 560B total parameter MoE model uses 25B active params, and is on par with other frontier open weights models like Kimi K3 and DeepSeek V4 Pro. https://t.co/uBQqb5vAE3
Jev classification details
Overall: Neutral68.0% confidence
The model is discussed in this dimension without a clear positive or negative opinion.
Confidence is Jev’s estimate, not verified accuracy. These are the saved classification results; Jev does not provide a written explanation for this post.
Overall label probabilities
Positive
23.0%
Negative
0.0%
Neutral
74.0%
Mixed
0.0%
Not discussed
3.0%
Cannot determine
0.0%
Jev’s category decisions
Dimension
Sentiment
Confidence
Overall
Neutral
68.0%
Reasoning
Not discussed
99.0%
Speed
Not discussed
100.0%
Cost
Not discussed
100.0%
Coding
Not discussed
99.0%
Refers to DeepSeek V4 ProYes · 63.0% confidence
jev-sentiment-v4/jev-1.13.0Classified Oct 5, 2026 · UTC
positiveJev confidence 100.0%
gotta be honest I think u guys are overhyping opus. spent $45 in tokens getting opus to build half of an accounting system and it ran a bunch of tests and burned credits at an exponential rate. Switched to DeepSeek v4 pro and it finished the rest of it for $1. what am I missing?
Jev classification details
Overall: Positive100.0% confidence
Praise, endorsement, or a favorable opinion about the model in this dimension.
Confidence is Jev’s estimate, not verified accuracy. These are the saved classification results; Jev does not provide a written explanation for this post.
Overall label probabilities
Positive
100.0%
Negative
0.0%
Neutral
0.0%
Mixed
0.0%
Not discussed
0.0%
Cannot determine
0.0%
Jev’s category decisions
Dimension
Sentiment
Confidence
Overall
Positive
100.0%
Reasoning
Not discussed
94.0%
Speed
Not discussed
99.0%
Cost
Positive
100.0%
Coding
Positive
99.0%
Refers to DeepSeek V4 ProYes · 99.0% confidence
jev-sentiment-v4/jev-1.13.0Classified Oct 5, 2026 · UTC
negativeJev confidence 82.0%
I just published the first model I've ever trained 🙌
#Qwen 3.8-27B fine-tuned on Solidity audit data. On 191 held-out contracts it scores +23% F1 over the base model, and even edges out #DeepSeek V4 Pro.
Learned a ton doing it. Weights, code and write-up 👇 https://t.co/fCYLaqXcyv
Jev classification details
Overall: Negative82.0% confidence
Criticism, disappointment, or an unfavorable opinion about the model in this dimension.
Confidence is Jev’s estimate, not verified accuracy. These are the saved classification results; Jev does not provide a written explanation for this post.
Overall label probabilities
Positive
3.0%
Negative
85.0%
Neutral
11.0%
Mixed
0.0%
Not discussed
1.0%
Cannot determine
0.0%
Jev’s category decisions
Dimension
Sentiment
Confidence
Overall
Negative
82.0%
Reasoning
Not discussed
94.0%
Speed
Not discussed
100.0%
Cost
Not discussed
100.0%
Coding
Negative
31.0%
Refers to DeepSeek V4 ProYes · 95.0% confidence
jev-sentiment-v4/jev-1.13.0Classified Oct 5, 2026 · UTC
Page 1 of 8 · 94 mentions
1 of 1 enabled models have a recorded collection outcome with target 50 in the last 7 days. Search exhaustion and page limits can produce fewer records.Target reached · 100 / 50 accepted · target reached · Last successful collection Oct 4, 2026
Last collection attempt Oct 5, 2026 · partially completedNew analyses use TypeSafe Jev