New engine releases & news H1 2026

[Elorejano]

Hi, Graham.
FireFly (if i remember ) needs some files that ar absent. Like an ini file, openings books etc
Do you have those ones?

It is a UCI engine, so the options will become available when loading it.

Its an XBOARD/uci engine, so those xboard files are needed. Im not use Uci protocol. Im using uci engines with polyglot under winboard

[Gabor Szots]

Hi, Graham.
FireFly (if i remember ) needs some files that ar absent. Like an ini file, openings books etc
Do you have those ones?

It is a UCI engine, so the options will become available when loading it.

Its an XBOARD/uci engine, so those xboard files are needed. Im not use Uci protocol. Im using uci engines with polyglot under winboard

It requires a file named FireFlyStart.txt. The contents of that file as I use it is as follows:

egtbpath F:\Sakk\TB
poshash 256MB
pawnhash 16MB
*ignoreucihash
*usebitbase

[Elorejano]

Hi, Graham.
FireFly (if i remember ) needs some files that ar absent. Like an ini file, openings books etc
Do you have those ones?

It is a UCI engine, so the options will become available when loading it.

Its an XBOARD/uci engine, so those xboard files are needed. Im not use Uci protocol. Im using uci engines with polyglot under winboard

It requires a file named FireFlyStart.txt. The contents of that file as I use it is as follows:

egtbpath F:\Sakk\TB
poshash 256MB
pawnhash 16MB
*ignoreucihash
*usebitbase

Thanks, Gabor.
And the opening book? If im remember well, FireFly use its own format

[DreamerExx]

Ember 1.1.0 has been released . What's improved?
1. My personal NNUE (100SB) trained by me
2. A huge number of bugs have been fixed compared to 1.0.0
3. The Elo fluctuates around 2850-2950 according to my calculations (1.0.0 was approximately 2750)
Download: https://github.com/ExxDreamerCode/Ember ... tag/V1.1.0

[NathanDrake]

Triumviratus 5.0 Released
https://github.com/Tors3/Triumviratus/releases/tag/v5.0

• Evaluation: NNUE, SFNNv13 architecture (Full_Threats + HalfKAv2_hm, threat-aware).
• Network: Own-lineage nn-rubicon-alea-v1, trained from scratch (no Stockfish network seed) using nnue-pytorch.
  • Stage 1: Stockfish 5000-node data (λ = 1.0 → 0.75).
  • Stage 2: Leela self-play (λ = 0.74, fixed).
• Search: Original SPSA-tuned alpha-beta search, Lazy SMP, Syzygy tablebases.
• Strength: Approximately +50 Elo over v4.2 (internal testing, 20+0.2).

I know 4.2 is still being tested, apologies for the overlap, but 5.0 was ready and I didn't want to sit on it.

[cc2150dx]

PlentyChess 8.0.0

- Introduction of the new threat inputs NNUE architecture, which has since been introduced in many other top engines (including Stockfish, Reckless and others). As of late, the NNUE additionally has pawn-pair inputs
- Many speedups to make the NNUE architecture changes viable
- Greatly improved search logic, pushing the concept of fractional depth further to more heuristics
- SMP improvements: Sharing correction histories between threads and tuning under multithreaded conditions
- Proper NUMA handling on linux
- General improvements to the source code

https://github.com/Yoshie2000/PlentyChe ... g/b-v8.0.0

[vihaa]

Hi,

I am pleased to announce the public release of my UCI chess engine MagicTree 3.2, written in C++.

Download / website: https://magictree.netlify.app/

MagicTree is a 64-bit Windows UCI engine. It is a classical handcrafted-evaluation engine, with no neural network, no NNUE file, and no external evaluation file included.

The engine uses bitboards, BMI2/PEXT-based sliding attack generation, alpha-beta / PVS search, iterative deepening, transposition tables, move-ordering heuristics, pruning, and a tapered handcrafted evaluation.

Important requirement: MagicTree 3.2 requires a 64-bit Windows system with a CPU supporting BMI2/PEXT instructions. On older CPUs without BMI2/PEXT support, this build may fail to start.

Estimated strength: ~2725 blitz, this is a private-test estimate until the engine receives more external testing.

Current limitations:
- Windows x64 only
- BMI2/PEXT-capable CPU required
- No opening book included
- No Syzygy/tablebase support included
- No NNUE or neural-network file included
- Single-threaded release

I have tested the engine mainly under Windows x64 in UCI-compatible GUIs. I would be grateful for any bug reports, stability issues, GUI compatibility feedback, tournament results, or suggestions for improvement.

Looking forward to seeing how MagicTree performs in external testing.

Regards,
Vikrant