DeepSeek V4 Flash 0731: 82.7% on Terminal-Bench 2.1 with a public harness

#1

Post by ubermon » Sun, Aug 09, 2026, 8:44 AM UTC

DeepSeek V4 Flash 0731: 82.7% on Terminal-Bench 2.1 with a public harness
antigma.ai

Re: DeepSeek V4 Flash 0731: 82.7% on Terminal-Bench 2.1 with a public harness

#2

Post by freakynit » Sun, Aug 09, 2026, 8:55 AM UTC

Keeps throwing forbidden after logging-in to the dashboard and goes back to login screen. Tried 4 times.

Re: DeepSeek V4 Flash 0731: 82.7% on Terminal-Bench 2.1 with a public harness

#3

Post by swrrt » Sun, Aug 09, 2026, 12:24 PM UTC

freakynit wrote: post #2

Keeps throwing forbidden after logging-in to the dashboard and goes back to login screen. Tried 4 times.

Oh, it seems fine for me. I can login actually. Maybe network/server instability?

Re: DeepSeek V4 Flash 0731: 82.7% on Terminal-Bench 2.1 with a public harness

#4

Post by freakynit » Sun, Aug 09, 2026, 4:38 PM UTC

swrrt wrote: post #3

freakynit wrote: post #2

Keeps throwing forbidden after logging-in to the dashboard and goes back to login screen. Tried 4 times.

Oh, it seems fine for me. I can login actually. Maybe network/server instability?

No idea..

Re: DeepSeek V4 Flash 0731: 82.7% on Terminal-Bench 2.1 with a public harness

#5

Post by kimjune01 » Sun, Aug 09, 2026, 6:23 PM UTC

an audit for the bench for the curious:

https://www.june.kim/terminal-bench-frame

Re: DeepSeek V4 Flash 0731: 82.7% on Terminal-Bench 2.1 with a public harness

#6

Post by ubermon » Sun, Aug 09, 2026, 7:13 PM UTC

kimjune01 wrote: post #5

an audit for the bench for the curious: https://www.june.kim/terminal-bench-frame

yeah, that bench has a lot of room to improve, but it is the best we can find now

Re: DeepSeek V4 Flash 0731: 82.7% on Terminal-Bench 2.1 with a public harness

#7

Post by polymer8563 » Mon, Aug 10, 2026, 1:21 AM UTC

From README for Ante:

"The core harness itself is developed in a private repository during the alpha"

lost all interest after that sentence, sad thing